Compare commits

..

No commits in common. "imports/c8-beta/device-mapper-multipath-0.8.4-22.el8" and "c8" have entirely different histories.

48 changed files with 6897 additions and 4 deletions

View File

@ -0,0 +1,31 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Xose Vazquez Perez <xose.vazquez@gmail.com>
Date: Wed, 9 Sep 2020 17:48:05 +0200
Subject: [PATCH] multipath-tools: add HPE MSA 1060/2060 to hwtable
Cc: Martin Wilck <mwilck@suse.com>
Cc: Christophe Varoqui <christophe.varoqui@opensvc.com>
Cc: DM-DEVEL ML <dm-devel@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/hwtable.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/libmultipath/hwtable.c b/libmultipath/hwtable.c
index d680bdfc..59bc0d6e 100644
--- a/libmultipath/hwtable.c
+++ b/libmultipath/hwtable.c
@@ -181,9 +181,9 @@ static struct hwentry default_hw[] = {
.prio_name = PRIO_ALUA,
},
{
- /* MSA 1040, 1050, 2040 and 2050 families */
+ /* MSA 1040, 1050, 1060, 2040, 2050 and 2060 families */
.vendor = "HP",
- .product = "MSA [12]0[45]0 SA[NS]",
+ .product = "MSA [12]0[456]0 SA[NS]",
.pgpolicy = GROUP_BY_PRIO,
.pgfailback = -FAILBACK_IMMEDIATE,
.no_path_retry = 18,

View File

@ -0,0 +1,78 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Uday Shankar <ushankar@purestorage.com>
Date: Wed, 9 Mar 2022 13:03:26 -0700
Subject: [PATCH] multipath-tools: update mpp->force_readonly in ev_add_path
When NVMe disks are added to the system, no uevent containing the
DISK_RO property is generated. As a result, dm-* nodes backed by
readonly NVMe disks will not have their RO state set properly. The
result looks like this:
$ multipath -l
eui.00c92c091fd6564424a9376600011bd1 dm-3 NVME,Pure Storage FlashArray
size=1.0T features='0' hwhandler='0' wp=rw
|-+- policy='service-time 0' prio=0 status=active
| `- 0:2:2:72657 nvme0n2 259:4 active undef running
`-+- policy='service-time 0' prio=0 status=enabled
`- 1:0:2:72657 nvme1n2 259:1 active undef running
$ cat /sys/block/dm-3/ro
0
$ cat /sys/block/nvme*n2/ro
1
1
This is not a problem for SCSI disks, since the kernel will emit change
uevents containing the DISK_RO property when the disk is added to the
system. See the following thread for my initial attempt to fix this
issue at the kernel level:
https://lore.kernel.org/linux-block/Yib8GqCA5e3SQYty@infradead.org/T/#t
Fix the issue by picking up the path ro state from sysfs in ev_add_path,
setting the mpp->force_readonly accordingly, and changing
dm_addmap_create to be aware of mpp->force_readonly.
Signed-off-by: Uday Shankar <ushankar@purestorage.com>
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/devmapper.c | 2 +-
multipathd/main.c | 6 ++++++
2 files changed, 7 insertions(+), 1 deletion(-)
diff --git a/libmultipath/devmapper.c b/libmultipath/devmapper.c
index 3f70e576..e3c3c119 100644
--- a/libmultipath/devmapper.c
+++ b/libmultipath/devmapper.c
@@ -420,7 +420,7 @@ int dm_addmap_create (struct multipath *mpp, char * params)
int ro;
uint16_t udev_flags = build_udev_flags(mpp, 0);
- for (ro = 0; ro <= 1; ro++) {
+ for (ro = mpp->force_readonly ? 1 : 0; ro <= 1; ro++) {
int err;
if (dm_addmap(DM_DEVICE_CREATE, TGT_MPATH, mpp, params, ro,
diff --git a/multipathd/main.c b/multipathd/main.c
index f4b79882..eeded52b 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -988,6 +988,7 @@ ev_add_path (struct path * pp, struct vectors * vecs, int need_do_map)
int retries = 3;
int start_waiter = 0;
int ret;
+ int ro;
/*
* need path UID to go any further
@@ -1051,6 +1052,11 @@ rescan:
/* persistent reservation check*/
mpath_pr_event_handle(pp);
+ /* ro check - if new path is ro, force map to be ro as well */
+ ro = sysfs_get_ro(pp);
+ if (ro == 1)
+ mpp->force_readonly = 1;
+
if (!need_do_map)
return 0;

View File

@ -0,0 +1,28 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 29 Apr 2022 15:57:12 -0500
Subject: [PATCH] updated HPE MSA builtin config
Make the config better align to MSA 4th, 5th and 6th Generation systems.
Cc: Jon.Paul@hpe.com
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/hwtable.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/libmultipath/hwtable.c b/libmultipath/hwtable.c
index 59bc0d6e..d6325864 100644
--- a/libmultipath/hwtable.c
+++ b/libmultipath/hwtable.c
@@ -182,8 +182,8 @@ static struct hwentry default_hw[] = {
},
{
/* MSA 1040, 1050, 1060, 2040, 2050 and 2060 families */
- .vendor = "HP",
- .product = "MSA [12]0[456]0 SA[NS]",
+ .vendor = "(HP|HPE)",
+ .product = "MSA [12]0[456]0 (SAN|SAS|FC|iSCSI)",
.pgpolicy = GROUP_BY_PRIO,
.pgfailback = -FAILBACK_IMMEDIATE,
.no_path_retry = 18,

View File

@ -0,0 +1,35 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 30 Mar 2022 15:14:56 -0500
Subject: [PATCH] multipath: return failure on an invalid remove command
When "multipath -f" is run on a device that doesn't exist or isn't a
multipath device, the command will not remove any device, but it will
still return success. Multiple functions rely on _dm_flush_map()
returning success when called with name that doesn't match any
multipath device. So before calling _dm_flush_map(), call dm_is_mpath(),
to check if the device exists and is a multipath device, and return
failure if it's not.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com
---
multipath/main.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/multipath/main.c b/multipath/main.c
index 14d045c9..d5e6499c 100644
--- a/multipath/main.c
+++ b/multipath/main.c
@@ -1150,6 +1150,11 @@ main (int argc, char *argv[])
if (retries < 0)
retries = conf->remove_retries;
if (cmd == CMD_FLUSH_ONE) {
+ if (dm_is_mpath(dev) != 1) {
+ condlog(0, "%s is not a multipath device", dev);
+ r = RTVL_FAIL;
+ goto out;
+ }
r = dm_suspend_and_flush_map(dev, retries) ?
RTVL_FAIL : RTVL_OK;
goto out;

View File

@ -0,0 +1,67 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:35 -0500
Subject: [PATCH] libmultipath: steal the src string pointer in merge_str()
Instead of allocating a copy when the original string is going to be
freed right after the merge, just steal the pointer. Also, merge_mpe()
can't get called with NULL arguments.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/config.c | 16 +++++-----------
1 file changed, 5 insertions(+), 11 deletions(-)
diff --git a/libmultipath/config.c b/libmultipath/config.c
index abbddaf1..aa79561e 100644
--- a/libmultipath/config.c
+++ b/libmultipath/config.c
@@ -323,9 +323,9 @@ set_param_str(const char * str)
}
#define merge_str(s) \
- if (!dst->s && src->s) { \
- if (!(dst->s = set_param_str(src->s))) \
- return 1; \
+ if (!dst->s && src->s && strlen(src->s)) { \
+ dst->s = src->s; \
+ src->s = NULL; \
}
#define merge_num(s) \
@@ -333,7 +333,7 @@ set_param_str(const char * str)
dst->s = src->s
-static int
+static void
merge_hwe (struct hwentry * dst, struct hwentry * src)
{
char id[SCSI_VENDOR_SIZE+PATH_PRODUCT_SIZE];
@@ -385,15 +385,11 @@ merge_hwe (struct hwentry * dst, struct hwentry * src)
reconcile_features_with_options(id, &dst->features,
&dst->no_path_retry,
&dst->retain_hwhandler);
- return 0;
}
-static int
+static void
merge_mpe(struct mpentry *dst, struct mpentry *src)
{
- if (!dst || !src)
- return 1;
-
merge_str(alias);
merge_str(uid_attribute);
merge_str(getuid);
@@ -435,8 +431,6 @@ merge_mpe(struct mpentry *dst, struct mpentry *src)
merge_num(uid);
merge_num(gid);
merge_num(mode);
-
- return 0;
}
void merge_mptable(vector mptable)

View File

@ -0,0 +1,75 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Thu, 17 Feb 2022 17:22:32 +0100
Subject: [PATCH] libmultipath: enable linear ordering of bus/proto tuple
We categorized protocols by bus/proto_id, while we only differentiate
protocol IDs for SCSI. Allow transforming this into a linear sequence
of bus/protocol IDs by having non-SCSI first, and follwing up with
the different SCSI protocols.
Signed-off-by: Martin Wilck <mwilck@suse.com>
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/structs.c | 10 ++++++++++
libmultipath/structs.h | 14 ++++++++++++--
2 files changed, 22 insertions(+), 2 deletions(-)
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index dda9884c..aaf85297 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -722,3 +722,13 @@ out:
return 0;
}
+
+unsigned int bus_protocol_id(const struct path *pp) {
+ if (!pp || pp->bus < 0 || pp->bus > SYSFS_BUS_SCSI)
+ return SYSFS_BUS_UNDEF;
+ if (pp->bus != SYSFS_BUS_SCSI)
+ return pp->bus;
+ if ((int)pp->sg_id.proto_id < 0 || pp->sg_id.proto_id > SCSI_PROTOCOL_UNSPEC)
+ return SYSFS_BUS_UNDEF;
+ return SYSFS_BUS_SCSI + pp->sg_id.proto_id;
+}
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index a5dbad5b..5e29ae38 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -57,12 +57,13 @@ enum failback_mode {
FAILBACK_FOLLOWOVER
};
+/* SYSFS_BUS_SCSI should be last, see bus_protocol_id() */
enum sysfs_buses {
SYSFS_BUS_UNDEF,
- SYSFS_BUS_SCSI,
SYSFS_BUS_CCW,
SYSFS_BUS_CCISS,
SYSFS_BUS_NVME,
+ SYSFS_BUS_SCSI,
};
enum pathstates {
@@ -190,9 +191,18 @@ enum scsi_protocol {
SCSI_PROTOCOL_SAS = 6,
SCSI_PROTOCOL_ADT = 7, /* Media Changers */
SCSI_PROTOCOL_ATA = 8,
- SCSI_PROTOCOL_UNSPEC = 0xf, /* No specific protocol */
+ SCSI_PROTOCOL_USB = 9, /* USB Attached SCSI (UAS), and others */
+ SCSI_PROTOCOL_UNSPEC = 0xa, /* No specific protocol */
};
+/*
+ * Linear ordering of bus/protocol
+ * This assumes that SYSFS_BUS_SCSI is last in enum sysfs_buses
+ * SCSI is the only bus type for which we distinguish protocols.
+ */
+#define LAST_BUS_PROTOCOL_ID (SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC)
+unsigned int bus_protocol_id(const struct path *pp);
+
enum no_undef_states {
NU_NO = -1,
NU_UNDEF = 0,

View File

@ -0,0 +1,82 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Thu, 17 Feb 2022 17:24:25 +0100
Subject: [PATCH] libmultipath: use bus_protocol_id() in
snprint_path_protocol()
Simplify bus_protocol_id() by using the linear ordering.
Signed-off-by: Martin Wilck <mwilck@suse.com>
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/print.c | 56 +++++++++++++++++---------------------------
1 file changed, 21 insertions(+), 35 deletions(-)
diff --git a/libmultipath/print.c b/libmultipath/print.c
index 298b3764..ff4d1854 100644
--- a/libmultipath/print.c
+++ b/libmultipath/print.c
@@ -662,41 +662,27 @@ snprint_path_failures(char * buff, size_t len, const struct path * pp)
int
snprint_path_protocol(char * buff, size_t len, const struct path * pp)
{
- switch (pp->bus) {
- case SYSFS_BUS_SCSI:
- switch (pp->sg_id.proto_id) {
- case SCSI_PROTOCOL_FCP:
- return snprintf(buff, len, "scsi:fcp");
- case SCSI_PROTOCOL_SPI:
- return snprintf(buff, len, "scsi:spi");
- case SCSI_PROTOCOL_SSA:
- return snprintf(buff, len, "scsi:ssa");
- case SCSI_PROTOCOL_SBP:
- return snprintf(buff, len, "scsi:sbp");
- case SCSI_PROTOCOL_SRP:
- return snprintf(buff, len, "scsi:srp");
- case SCSI_PROTOCOL_ISCSI:
- return snprintf(buff, len, "scsi:iscsi");
- case SCSI_PROTOCOL_SAS:
- return snprintf(buff, len, "scsi:sas");
- case SCSI_PROTOCOL_ADT:
- return snprintf(buff, len, "scsi:adt");
- case SCSI_PROTOCOL_ATA:
- return snprintf(buff, len, "scsi:ata");
- case SCSI_PROTOCOL_UNSPEC:
- default:
- return snprintf(buff, len, "scsi:unspec");
- }
- case SYSFS_BUS_CCW:
- return snprintf(buff, len, "ccw");
- case SYSFS_BUS_CCISS:
- return snprintf(buff, len, "cciss");
- case SYSFS_BUS_NVME:
- return snprintf(buff, len, "nvme");
- case SYSFS_BUS_UNDEF:
- default:
- return snprintf(buff, len, "undef");
- }
+ static const char * const protocol_name[LAST_BUS_PROTOCOL_ID + 1] = {
+ [SYSFS_BUS_UNDEF] = "undef",
+ [SYSFS_BUS_CCW] = "ccw",
+ [SYSFS_BUS_CCISS] = "cciss",
+ [SYSFS_BUS_NVME] = "nvme",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_FCP] = "scsi:fcp",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SPI] = "scsi:spi",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SSA] = "scsi:ssa",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SBP] = "scsi:sbp",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SRP] = "scsi:srp",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ISCSI] = "scsi:iscsi",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SAS] = "scsi:sas",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ADT] = "scsi:adt",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ATA] = "scsi:ata",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_USB] = "scsi:usb",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC] = "scsi:unspec",
+ };
+ const char *pn = protocol_name[bus_protocol_id(pp)];
+
+ assert(pn != NULL);
+ return snprintf(buff, len, "%s", pn);
}
int

View File

@ -0,0 +1,85 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:36 -0500
Subject: [PATCH] libmultipath: make protocol_name global
Future patches will make use of this, so move it out of
snprint_path_protocol()
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/print.c | 17 -----------------
libmultipath/structs.c | 18 ++++++++++++++++++
libmultipath/structs.h | 1 +
3 files changed, 19 insertions(+), 17 deletions(-)
diff --git a/libmultipath/print.c b/libmultipath/print.c
index ff4d1854..1f6d27bd 100644
--- a/libmultipath/print.c
+++ b/libmultipath/print.c
@@ -662,23 +662,6 @@ snprint_path_failures(char * buff, size_t len, const struct path * pp)
int
snprint_path_protocol(char * buff, size_t len, const struct path * pp)
{
- static const char * const protocol_name[LAST_BUS_PROTOCOL_ID + 1] = {
- [SYSFS_BUS_UNDEF] = "undef",
- [SYSFS_BUS_CCW] = "ccw",
- [SYSFS_BUS_CCISS] = "cciss",
- [SYSFS_BUS_NVME] = "nvme",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_FCP] = "scsi:fcp",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SPI] = "scsi:spi",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SSA] = "scsi:ssa",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SBP] = "scsi:sbp",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SRP] = "scsi:srp",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ISCSI] = "scsi:iscsi",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SAS] = "scsi:sas",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ADT] = "scsi:adt",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ATA] = "scsi:ata",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_USB] = "scsi:usb",
- [SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC] = "scsi:unspec",
- };
const char *pn = protocol_name[bus_protocol_id(pp)];
assert(pn != NULL);
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index aaf85297..19099bed 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -20,6 +20,24 @@
#include "prioritizers/alua_spc3.h"
#include "dm-generic.h"
+const char * const protocol_name[LAST_BUS_PROTOCOL_ID + 1] = {
+ [SYSFS_BUS_UNDEF] = "undef",
+ [SYSFS_BUS_CCW] = "ccw",
+ [SYSFS_BUS_CCISS] = "cciss",
+ [SYSFS_BUS_NVME] = "nvme",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_FCP] = "scsi:fcp",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SPI] = "scsi:spi",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SSA] = "scsi:ssa",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SBP] = "scsi:sbp",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SRP] = "scsi:srp",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ISCSI] = "scsi:iscsi",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_SAS] = "scsi:sas",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ADT] = "scsi:adt",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_ATA] = "scsi:ata",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_USB] = "scsi:usb",
+ [SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC] = "scsi:unspec",
+};
+
struct adapter_group *
alloc_adaptergroup(void)
{
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 5e29ae38..ab99852f 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -202,6 +202,7 @@ enum scsi_protocol {
*/
#define LAST_BUS_PROTOCOL_ID (SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC)
unsigned int bus_protocol_id(const struct path *pp);
+extern const char * const protocol_name[];
enum no_undef_states {
NU_NO = -1,

View File

@ -0,0 +1,419 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:37 -0500
Subject: [PATCH] libmultipath: add a protocol subsection to multipath.conf
Some storage arrays can be accessed using multiple protocols at the same
time. In these cases, users may want to set path attributes
differently, depending on the protocol that the path is using. To allow
this, add a protocol subsection to the overrides section in
multipath.conf, which allows select path-specific options to be set.
This commit simply adds the subsection, and handles merging matching
entries. Future patches will make use of the section.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/config.c | 83 ++++++++++++++++++++++++++++++++++++
libmultipath/config.h | 10 +++++
libmultipath/dict.c | 99 +++++++++++++++++++++++++++++++++++++++++++
libmultipath/print.c | 44 +++++++++++++++++++
4 files changed, 236 insertions(+)
diff --git a/libmultipath/config.c b/libmultipath/config.c
index aa79561e..88975323 100644
--- a/libmultipath/config.c
+++ b/libmultipath/config.c
@@ -173,6 +173,18 @@ char *get_mpe_wwid(vector mptable, char *alias)
return NULL;
}
+static void
+free_pctable (vector pctable)
+{
+ int i;
+ struct pcentry *pce;
+
+ vector_foreach_slot(pctable, pce, i)
+ free(pce);
+
+ vector_free(pctable);
+}
+
void
free_hwe (struct hwentry * hwe)
{
@@ -218,6 +230,9 @@ free_hwe (struct hwentry * hwe)
if (hwe->bl_product)
FREE(hwe->bl_product);
+ if (hwe->pctable)
+ free_pctable(hwe->pctable);
+
FREE(hwe);
}
@@ -299,6 +314,15 @@ alloc_hwe (void)
return hwe;
}
+struct pcentry *
+alloc_pce (void)
+{
+ struct pcentry *pce = (struct pcentry *)
+ calloc(1, sizeof(struct pcentry));
+ pce->type = -1;
+ return pce;
+}
+
static char *
set_param_str(const char * str)
{
@@ -332,6 +356,13 @@ set_param_str(const char * str)
if (!dst->s && src->s) \
dst->s = src->s
+static void
+merge_pce(struct pcentry *dst, struct pcentry *src)
+{
+ merge_num(fast_io_fail);
+ merge_num(dev_loss);
+ merge_num(eh_deadline);
+}
static void
merge_hwe (struct hwentry * dst, struct hwentry * src)
@@ -538,6 +569,51 @@ out:
return 1;
}
+static void
+validate_pctable(struct hwentry *ovr, int idx, const char *table_desc)
+{
+ struct pcentry *pce;
+
+ if (!ovr || !ovr->pctable)
+ return;
+
+ vector_foreach_slot_after(ovr->pctable, pce, idx) {
+ if (pce->type < 0) {
+ condlog(0, "protocol section in %s missing type",
+ table_desc);
+ vector_del_slot(ovr->pctable, idx--);
+ free(pce);
+ }
+ }
+
+ if (VECTOR_SIZE(ovr->pctable) == 0) {
+ vector_free(ovr->pctable);
+ ovr->pctable = NULL;
+ }
+}
+
+static void
+merge_pctable(struct hwentry *ovr)
+{
+ struct pcentry *pce1, *pce2;
+ int i, j;
+
+ if (!ovr || !ovr->pctable)
+ return;
+
+ vector_foreach_slot(ovr->pctable, pce1, i) {
+ j = i + 1;
+ vector_foreach_slot_after(ovr->pctable, pce2, j) {
+ if (pce1->type != pce2->type)
+ continue;
+ merge_pce(pce2,pce1);
+ vector_del_slot(ovr->pctable, i--);
+ free(pce1);
+ break;
+ }
+ }
+}
+
static void
factorize_hwtable (vector hw, int n, const char *table_desc)
{
@@ -666,6 +742,7 @@ process_config_dir(struct config *conf, char *dir)
int i, n;
char path[LINE_MAX];
int old_hwtable_size;
+ int old_pctable_size = 0;
if (dir[0] != '/') {
condlog(1, "config_dir '%s' must be a fully qualified path",
@@ -692,11 +769,15 @@ process_config_dir(struct config *conf, char *dir)
continue;
old_hwtable_size = VECTOR_SIZE(conf->hwtable);
+ old_pctable_size = conf->overrides ?
+ VECTOR_SIZE(conf->overrides->pctable) : 0;
snprintf(path, LINE_MAX, "%s/%s", dir, namelist[i]->d_name);
path[LINE_MAX-1] = '\0';
process_file(conf, path);
factorize_hwtable(conf->hwtable, old_hwtable_size,
namelist[i]->d_name);
+ validate_pctable(conf->overrides, old_pctable_size,
+ namelist[i]->d_name);
}
pthread_cleanup_pop(1);
}
@@ -784,6 +865,7 @@ load_config (char * file)
goto out;
}
factorize_hwtable(conf->hwtable, builtin_hwtable_size, file);
+ validate_pctable(conf->overrides, 0, file);
} else {
condlog(0, "/etc/multipath.conf does not exist, blacklisting all devices.");
condlog(0, "You can run \"/sbin/mpathconf --enable\" to create");
@@ -898,6 +980,7 @@ load_config (char * file)
goto out;
}
+ merge_pctable(conf->overrides);
merge_mptable(conf->mptable);
merge_blacklist(conf->blist_devnode);
merge_blacklist(conf->blist_property);
diff --git a/libmultipath/config.h b/libmultipath/config.h
index e2e3f143..143116b3 100644
--- a/libmultipath/config.h
+++ b/libmultipath/config.h
@@ -41,6 +41,13 @@ enum force_reload_types {
FORCE_RELOAD_WEAK,
};
+struct pcentry {
+ int type;
+ int fast_io_fail;
+ unsigned int dev_loss;
+ int eh_deadline;
+};
+
struct hwentry {
char * vendor;
char * product;
@@ -86,6 +93,8 @@ struct hwentry {
int vpd_vendor_id;
int recheck_wwid;
char * bl_product;
+
+ vector pctable;
};
struct mpentry {
@@ -240,6 +249,7 @@ char * get_mpe_wwid (vector mptable, char * alias);
struct hwentry * alloc_hwe (void);
struct mpentry * alloc_mpe (void);
+struct pcentry * alloc_pce (void);
void free_hwe (struct hwentry * hwe);
void free_hwtable (vector hwtable);
diff --git a/libmultipath/dict.c b/libmultipath/dict.c
index 5a0255b0..8321ec1e 100644
--- a/libmultipath/dict.c
+++ b/libmultipath/dict.c
@@ -457,6 +457,29 @@ snprint_mp_ ## option (struct config *conf, char * buff, int len, \
return function (buff, len, mpe->option); \
}
+#define declare_pc_handler(option, function) \
+static int \
+pc_ ## option ## _handler (struct config *conf, vector strvec, \
+ const char *file, int line_nr) \
+{ \
+ struct pcentry *pce; \
+ if (!conf->overrides || !conf->overrides->pctable) \
+ return 1; \
+ pce = VECTOR_LAST_SLOT(conf->overrides->pctable); \
+ if (!pce) \
+ return 1; \
+ return function (strvec, &pce->option, file, line_nr); \
+}
+
+#define declare_pc_snprint(option, function) \
+static int \
+snprint_pc_ ## option (struct config *conf, char * buff, int len, \
+ const void *data) \
+{ \
+ const struct pcentry *pce = (const struct pcentry *)data; \
+ return function(buff, len, pce->option); \
+}
+
static int checkint_handler(struct config *conf, vector strvec,
const char *file, int line_nr)
{
@@ -1022,6 +1045,8 @@ declare_ovr_handler(fast_io_fail, set_undef_off_zero)
declare_ovr_snprint(fast_io_fail, print_undef_off_zero)
declare_hw_handler(fast_io_fail, set_undef_off_zero)
declare_hw_snprint(fast_io_fail, print_undef_off_zero)
+declare_pc_handler(fast_io_fail, set_undef_off_zero)
+declare_pc_snprint(fast_io_fail, print_undef_off_zero)
static int
set_dev_loss(vector strvec, void *ptr, const char *file, int line_nr)
@@ -1059,6 +1084,8 @@ declare_ovr_handler(dev_loss, set_dev_loss)
declare_ovr_snprint(dev_loss, print_dev_loss)
declare_hw_handler(dev_loss, set_dev_loss)
declare_hw_snprint(dev_loss, print_dev_loss)
+declare_pc_handler(dev_loss, set_dev_loss)
+declare_pc_snprint(dev_loss, print_dev_loss)
declare_def_handler(eh_deadline, set_undef_off_zero)
declare_def_snprint(eh_deadline, print_undef_off_zero)
@@ -1066,6 +1093,8 @@ declare_ovr_handler(eh_deadline, set_undef_off_zero)
declare_ovr_snprint(eh_deadline, print_undef_off_zero)
declare_hw_handler(eh_deadline, set_undef_off_zero)
declare_hw_snprint(eh_deadline, print_undef_off_zero)
+declare_pc_handler(eh_deadline, set_undef_off_zero)
+declare_pc_snprint(eh_deadline, print_undef_off_zero)
static int
set_pgpolicy(vector strvec, void *ptr, const char *file, int line_nr)
@@ -1876,6 +1905,69 @@ declare_mp_snprint(wwid, print_str)
declare_mp_handler(alias, set_str_noslash)
declare_mp_snprint(alias, print_str)
+
+static int
+protocol_handler(struct config *conf, vector strvec, const char *file,
+ int line_nr)
+{
+ struct pcentry *pce;
+
+ if (!conf->overrides)
+ return 1;
+
+ if (!conf->overrides->pctable &&
+ !(conf->overrides->pctable = vector_alloc()))
+ return 1;
+
+ if (!(pce = alloc_pce()))
+ return 1;
+
+ if (!vector_alloc_slot(conf->overrides->pctable)) {
+ free(pce);
+ return 1;
+ }
+ vector_set_slot(conf->overrides->pctable, pce);
+
+ return 0;
+}
+
+static int
+set_protocol_type(vector strvec, void *ptr, const char *file, int line_nr)
+{
+ int *int_ptr = (int *)ptr;
+ char *buff;
+ int i;
+
+ buff = set_value(strvec);
+
+ if (!buff)
+ return 1;
+
+ for (i = 0; i <= LAST_BUS_PROTOCOL_ID; i++) {
+ if (protocol_name[i] && !strcmp(buff, protocol_name[i])) {
+ *int_ptr = i;
+ break;
+ }
+ }
+ if (i > LAST_BUS_PROTOCOL_ID)
+ condlog(1, "%s line %d, invalid value for type: \"%s\"",
+ file, line_nr, buff);
+
+ free(buff);
+ return 0;
+}
+
+static int
+print_protocol_type(char *buff, int len, int type)
+{
+ if (type < 0)
+ return 0;
+ return snprintf(buff, len, "\"%s\"", protocol_name[type]);
+}
+
+declare_pc_handler(type, set_protocol_type)
+declare_pc_snprint(type, print_protocol_type)
+
/*
* deprecated handlers
*/
@@ -2117,6 +2209,13 @@ init_keywords(vector keywords)
install_keyword("ghost_delay", &ovr_ghost_delay_handler, &snprint_ovr_ghost_delay);
install_keyword("all_tg_pt", &ovr_all_tg_pt_handler, &snprint_ovr_all_tg_pt);
install_keyword("recheck_wwid", &ovr_recheck_wwid_handler, &snprint_ovr_recheck_wwid);
+ install_keyword_multi("protocol", &protocol_handler, NULL);
+ install_sublevel();
+ install_keyword("type", &pc_type_handler, &snprint_pc_type);
+ install_keyword("fast_io_fail_tmo", &pc_fast_io_fail_handler, &snprint_pc_fast_io_fail);
+ install_keyword("dev_loss_tmo", &pc_dev_loss_handler, &snprint_pc_dev_loss);
+ install_keyword("eh_deadline", &pc_eh_deadline_handler, &snprint_pc_eh_deadline);
+ install_sublevel_end();
install_keyword_root("multipaths", &multipaths_handler);
install_keyword_multi("multipath", &multipath_handler, NULL);
diff --git a/libmultipath/print.c b/libmultipath/print.c
index 1f6d27bd..8a6fbe83 100644
--- a/libmultipath/print.c
+++ b/libmultipath/print.c
@@ -1392,6 +1392,39 @@ snprint_multipath_topology_json (char * buff, int len, const struct vectors * ve
return fwd;
}
+static int
+snprint_pcentry (const struct config *conf, char *buff, int len,
+ const struct pcentry *pce)
+{
+ int i;
+ int fwd = 0;
+ struct keyword *kw;
+ struct keyword *rootkw;
+
+ rootkw = find_keyword(conf->keywords, NULL, "overrides");
+ if (!rootkw || !rootkw->sub)
+ return 0;
+
+ rootkw = find_keyword(conf->keywords, rootkw->sub, "protocol");
+ if (!rootkw)
+ return 0;
+
+ fwd += snprintf(buff + fwd, len - fwd, "\tprotocol {\n");
+ if (fwd >= len)
+ return len;
+
+ iterate_sub_keywords(rootkw, kw, i) {
+ fwd += snprint_keyword(buff + fwd, len - fwd, "\t\t%k %v\n",
+ kw, pce);
+ if (fwd >= len)
+ return len;
+ }
+ fwd += snprintf(buff + fwd, len - fwd, "\t}\n");
+ if (fwd >= len)
+ return len;
+ return fwd;
+}
+
static int
snprint_hwentry (const struct config *conf,
char * buff, int len, const struct hwentry * hwe)
@@ -1575,6 +1608,17 @@ static int snprint_overrides(const struct config *conf, char * buff, int len,
if (fwd >= len)
return len;
}
+
+ if (overrides->pctable) {
+ struct pcentry *pce;
+
+ vector_foreach_slot(overrides->pctable, pce, i) {
+ fwd += snprint_pcentry(conf, buff + fwd, len - fwd,
+ pce);
+ if (fwd >= len)
+ return len;
+ }
+ }
out:
fwd += snprintf(buff + fwd, len - fwd, "}\n");
if (fwd >= len)

View File

@ -0,0 +1,472 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:38 -0500
Subject: [PATCH] libmultipath: Set the scsi timeout parameters by path
Instead of dev_loss, fast_io_fail, and eh_deadline belonging to the
multipath structure, have them belong to the path structure. This means
that they are selected per path, and that sysfs_set_scsi_tmo() doesn't
assume that all paths of a multipath device will have the same value.
Currently they will all be the same, but a future patch will make it
possible for paths to have different values based on their protocol.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/configure.c | 5 +-
libmultipath/discovery.c | 149 +++++++++++++++++++++++----------------
libmultipath/discovery.h | 2 +-
libmultipath/propsel.c | 42 +++++------
libmultipath/propsel.h | 6 +-
libmultipath/structs.c | 1 -
libmultipath/structs.h | 6 +-
7 files changed, 118 insertions(+), 93 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 9a9890f5..6cad0468 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -338,9 +338,6 @@ int setup_map(struct multipath *mpp, char *params, int params_size,
select_mode(conf, mpp);
select_uid(conf, mpp);
select_gid(conf, mpp);
- select_fast_io_fail(conf, mpp);
- select_dev_loss(conf, mpp);
- select_eh_deadline(conf, mpp);
select_reservation_key(conf, mpp);
select_deferred_remove(conf, mpp);
select_marginal_path_err_sample_time(conf, mpp);
@@ -356,7 +353,7 @@ int setup_map(struct multipath *mpp, char *params, int params_size,
select_ghost_delay(conf, mpp);
select_flush_on_last_del(conf, mpp);
- sysfs_set_scsi_tmo(mpp, conf->checkint);
+ sysfs_set_scsi_tmo(conf, mpp);
marginal_pathgroups = conf->marginal_pathgroups;
pthread_cleanup_pop(1);
diff --git a/libmultipath/discovery.c b/libmultipath/discovery.c
index 2404cb87..36cc389e 100644
--- a/libmultipath/discovery.c
+++ b/libmultipath/discovery.c
@@ -581,13 +581,13 @@ sysfs_get_asymmetric_access_state(struct path *pp, char *buff, int buflen)
}
static int
-sysfs_set_eh_deadline(struct multipath *mpp, struct path *pp)
+sysfs_set_eh_deadline(struct path *pp)
{
struct udev_device *hostdev;
char host_name[HOST_NAME_LEN], value[16];
int ret;
- if (mpp->eh_deadline == EH_DEADLINE_UNSET)
+ if (pp->eh_deadline == EH_DEADLINE_UNSET)
return 0;
sprintf(host_name, "host%d", pp->sg_id.host_no);
@@ -596,12 +596,12 @@ sysfs_set_eh_deadline(struct multipath *mpp, struct path *pp)
if (!hostdev)
return 1;
- if (mpp->eh_deadline == EH_DEADLINE_OFF)
+ if (pp->eh_deadline == EH_DEADLINE_OFF)
sprintf(value, "off");
- else if (mpp->eh_deadline == EH_DEADLINE_ZERO)
+ else if (pp->eh_deadline == EH_DEADLINE_ZERO)
sprintf(value, "0");
else
- snprintf(value, 16, "%u", mpp->eh_deadline);
+ snprintf(value, 16, "%u", pp->eh_deadline);
ret = sysfs_attr_set_value(hostdev, "eh_deadline",
value, strlen(value));
@@ -625,6 +625,9 @@ sysfs_set_rport_tmo(struct multipath *mpp, struct path *pp)
unsigned int tmo;
int ret;
+ if (!pp->dev_loss && pp->fast_io_fail == MP_FAST_IO_FAIL_UNSET)
+ return;
+
sprintf(rport_id, "rport-%d:%d-%d",
pp->sg_id.host_no, pp->sg_id.channel, pp->sg_id.transport_id);
rport_dev = udev_device_new_from_subsystem_sysname(udev,
@@ -664,14 +667,14 @@ sysfs_set_rport_tmo(struct multipath *mpp, struct path *pp)
* then set fast_io_fail, and _then_ set dev_loss_tmo
* to the correct value.
*/
- if (mpp->fast_io_fail != MP_FAST_IO_FAIL_UNSET &&
- mpp->fast_io_fail != MP_FAST_IO_FAIL_ZERO &&
- mpp->fast_io_fail != MP_FAST_IO_FAIL_OFF) {
+ if (pp->fast_io_fail != MP_FAST_IO_FAIL_UNSET &&
+ pp->fast_io_fail != MP_FAST_IO_FAIL_ZERO &&
+ pp->fast_io_fail != MP_FAST_IO_FAIL_OFF) {
/* Check if we need to temporarily increase dev_loss_tmo */
- if ((unsigned int)mpp->fast_io_fail >= tmo) {
+ if ((unsigned int)pp->fast_io_fail >= tmo) {
/* Increase dev_loss_tmo temporarily */
snprintf(value, sizeof(value), "%u",
- (unsigned int)mpp->fast_io_fail + 1);
+ (unsigned int)pp->fast_io_fail + 1);
ret = sysfs_attr_set_value(rport_dev, "dev_loss_tmo",
value, strlen(value));
if (ret <= 0) {
@@ -685,20 +688,20 @@ sysfs_set_rport_tmo(struct multipath *mpp, struct path *pp)
goto out;
}
}
- } else if (mpp->dev_loss > DEFAULT_DEV_LOSS_TMO &&
- mpp->no_path_retry != NO_PATH_RETRY_QUEUE) {
+ } else if (pp->dev_loss > DEFAULT_DEV_LOSS_TMO &&
+ mpp->no_path_retry != NO_PATH_RETRY_QUEUE) {
condlog(3, "%s: limiting dev_loss_tmo to %d, since "
"fast_io_fail is not set",
rport_id, DEFAULT_DEV_LOSS_TMO);
- mpp->dev_loss = DEFAULT_DEV_LOSS_TMO;
+ pp->dev_loss = DEFAULT_DEV_LOSS_TMO;
}
- if (mpp->fast_io_fail != MP_FAST_IO_FAIL_UNSET) {
- if (mpp->fast_io_fail == MP_FAST_IO_FAIL_OFF)
+ if (pp->fast_io_fail != MP_FAST_IO_FAIL_UNSET) {
+ if (pp->fast_io_fail == MP_FAST_IO_FAIL_OFF)
sprintf(value, "off");
- else if (mpp->fast_io_fail == MP_FAST_IO_FAIL_ZERO)
+ else if (pp->fast_io_fail == MP_FAST_IO_FAIL_ZERO)
sprintf(value, "0");
else
- snprintf(value, 16, "%u", mpp->fast_io_fail);
+ snprintf(value, 16, "%u", pp->fast_io_fail);
ret = sysfs_attr_set_value(rport_dev, "fast_io_fail_tmo",
value, strlen(value));
if (ret <= 0) {
@@ -709,8 +712,8 @@ sysfs_set_rport_tmo(struct multipath *mpp, struct path *pp)
rport_id, value, -ret);
}
}
- if (mpp->dev_loss > 0) {
- snprintf(value, 16, "%u", mpp->dev_loss);
+ if (pp->dev_loss > 0) {
+ snprintf(value, 16, "%u", pp->dev_loss);
ret = sysfs_attr_set_value(rport_dev, "dev_loss_tmo",
value, strlen(value));
if (ret <= 0) {
@@ -726,7 +729,7 @@ out:
}
static void
-sysfs_set_session_tmo(struct multipath *mpp, struct path *pp)
+sysfs_set_session_tmo(struct path *pp)
{
struct udev_device *session_dev = NULL;
char session_id[64];
@@ -743,18 +746,18 @@ sysfs_set_session_tmo(struct multipath *mpp, struct path *pp)
condlog(4, "target%d:%d:%d -> %s", pp->sg_id.host_no,
pp->sg_id.channel, pp->sg_id.scsi_id, session_id);
- if (mpp->dev_loss) {
+ if (pp->dev_loss) {
condlog(3, "%s: ignoring dev_loss_tmo on iSCSI", pp->dev);
}
- if (mpp->fast_io_fail != MP_FAST_IO_FAIL_UNSET) {
- if (mpp->fast_io_fail == MP_FAST_IO_FAIL_OFF) {
+ if (pp->fast_io_fail != MP_FAST_IO_FAIL_UNSET) {
+ if (pp->fast_io_fail == MP_FAST_IO_FAIL_OFF) {
condlog(3, "%s: can't switch off fast_io_fail_tmo "
"on iSCSI", pp->dev);
- } else if (mpp->fast_io_fail == MP_FAST_IO_FAIL_ZERO) {
+ } else if (pp->fast_io_fail == MP_FAST_IO_FAIL_ZERO) {
condlog(3, "%s: can't set fast_io_fail_tmo to '0'"
"on iSCSI", pp->dev);
} else {
- snprintf(value, 11, "%u", mpp->fast_io_fail);
+ snprintf(value, 11, "%u", pp->fast_io_fail);
if (sysfs_attr_set_value(session_dev, "recovery_tmo",
value, strlen(value)) <= 0) {
condlog(3, "%s: Failed to set recovery_tmo, "
@@ -767,12 +770,15 @@ sysfs_set_session_tmo(struct multipath *mpp, struct path *pp)
}
static void
-sysfs_set_nexus_loss_tmo(struct multipath *mpp, struct path *pp)
+sysfs_set_nexus_loss_tmo(struct path *pp)
{
struct udev_device *sas_dev = NULL;
char end_dev_id[64];
char value[11];
+ if (!pp->udev || !pp->dev_loss)
+ return;
+
sprintf(end_dev_id, "end_device-%d:%d",
pp->sg_id.host_no, pp->sg_id.transport_id);
sas_dev = udev_device_new_from_subsystem_sysname(udev,
@@ -785,8 +791,8 @@ sysfs_set_nexus_loss_tmo(struct multipath *mpp, struct path *pp)
condlog(4, "target%d:%d:%d -> %s", pp->sg_id.host_no,
pp->sg_id.channel, pp->sg_id.scsi_id, end_dev_id);
- if (mpp->dev_loss) {
- snprintf(value, 11, "%u", mpp->dev_loss);
+ if (pp->dev_loss) {
+ snprintf(value, 11, "%u", pp->dev_loss);
if (sysfs_attr_set_value(sas_dev, "I_T_nexus_loss_timeout",
value, strlen(value)) <= 0)
condlog(3, "%s: failed to update "
@@ -798,53 +804,76 @@ sysfs_set_nexus_loss_tmo(struct multipath *mpp, struct path *pp)
}
int
-sysfs_set_scsi_tmo (struct multipath *mpp, unsigned int checkint)
+sysfs_set_scsi_tmo (struct config *conf, struct multipath *mpp)
{
struct path *pp;
int i;
- unsigned int dev_loss_tmo = mpp->dev_loss;
+ unsigned int min_dev_loss = 0;
+ bool warn_dev_loss = false;
+ bool warn_fast_io_fail = false;
if (mpp->no_path_retry > 0) {
uint64_t no_path_retry_tmo =
- (uint64_t)mpp->no_path_retry * checkint;
+ (uint64_t)mpp->no_path_retry * conf->checkint;
if (no_path_retry_tmo > MAX_DEV_LOSS_TMO)
- no_path_retry_tmo = MAX_DEV_LOSS_TMO;
- if (no_path_retry_tmo > dev_loss_tmo)
- dev_loss_tmo = no_path_retry_tmo;
- condlog(3, "%s: update dev_loss_tmo to %u",
- mpp->alias, dev_loss_tmo);
- } else if (mpp->no_path_retry == NO_PATH_RETRY_QUEUE) {
- dev_loss_tmo = MAX_DEV_LOSS_TMO;
- condlog(3, "%s: update dev_loss_tmo to %u",
- mpp->alias, dev_loss_tmo);
- }
- mpp->dev_loss = dev_loss_tmo;
- if (mpp->dev_loss && mpp->fast_io_fail > 0 &&
- (unsigned int)mpp->fast_io_fail >= mpp->dev_loss) {
- condlog(3, "%s: turning off fast_io_fail (%d is not smaller than dev_loss_tmo)",
- mpp->alias, mpp->fast_io_fail);
- mpp->fast_io_fail = MP_FAST_IO_FAIL_OFF;
- }
- if (!mpp->dev_loss && mpp->fast_io_fail == MP_FAST_IO_FAIL_UNSET &&
- mpp->eh_deadline == EH_DEADLINE_UNSET)
- return 0;
+ min_dev_loss = MAX_DEV_LOSS_TMO;
+ else
+ min_dev_loss = no_path_retry_tmo;
+ } else if (mpp->no_path_retry == NO_PATH_RETRY_QUEUE)
+ min_dev_loss = MAX_DEV_LOSS_TMO;
vector_foreach_slot(mpp->paths, pp, i) {
+ select_fast_io_fail(conf, pp);
+ select_dev_loss(conf, pp);
+ select_eh_deadline(conf, pp);
+
+ if (!pp->dev_loss && pp->eh_deadline == EH_DEADLINE_UNSET &&
+ pp->fast_io_fail == MP_FAST_IO_FAIL_UNSET)
+ continue;
+
if (pp->bus != SYSFS_BUS_SCSI)
continue;
+ sysfs_set_eh_deadline(pp);
+
+ if (!pp->dev_loss && pp->fast_io_fail == MP_FAST_IO_FAIL_UNSET)
+ continue;
+
+ if (pp->sg_id.proto_id != SCSI_PROTOCOL_FCP &&
+ pp->sg_id.proto_id != SCSI_PROTOCOL_ISCSI &&
+ pp->sg_id.proto_id != SCSI_PROTOCOL_SAS)
+ continue;
+
+ if (pp->dev_loss > 0 && pp->dev_loss < min_dev_loss) {
+ warn_dev_loss = true;
+ pp->dev_loss = min_dev_loss;
+ }
+ if (pp->dev_loss > 0 && pp->fast_io_fail > 0 &&
+ (unsigned int)pp->fast_io_fail >= pp->dev_loss) {
+ warn_fast_io_fail = true;
+ pp->fast_io_fail = MP_FAST_IO_FAIL_OFF;
+ }
- if (mpp->dev_loss ||
- mpp->fast_io_fail != MP_FAST_IO_FAIL_UNSET) {
- if (pp->sg_id.proto_id == SCSI_PROTOCOL_FCP)
- sysfs_set_rport_tmo(mpp, pp);
- else if (pp->sg_id.proto_id == SCSI_PROTOCOL_ISCSI)
- sysfs_set_session_tmo(mpp, pp);
- else if (pp->sg_id.proto_id == SCSI_PROTOCOL_SAS)
- sysfs_set_nexus_loss_tmo(mpp, pp);
+ switch (pp->sg_id.proto_id) {
+ case SCSI_PROTOCOL_FCP:
+ sysfs_set_rport_tmo(mpp, pp);
+ break;
+ case SCSI_PROTOCOL_ISCSI:
+ sysfs_set_session_tmo(pp);
+ break;
+ case SCSI_PROTOCOL_SAS:
+ sysfs_set_nexus_loss_tmo(pp);
+ break;
+ default:
+ break;
}
- sysfs_set_eh_deadline(mpp, pp);
}
+ if (warn_dev_loss)
+ condlog(2, "%s: Raising dev_loss_tmo to %u because of no_path_retry setting",
+ mpp->alias, min_dev_loss);
+ if (warn_fast_io_fail)
+ condlog(3, "%s: turning off fast_io_fail (not smaller than dev_loss_tmo)",
+ mpp->alias);
return 0;
}
diff --git a/libmultipath/discovery.h b/libmultipath/discovery.h
index a5446b4d..b6eea258 100644
--- a/libmultipath/discovery.h
+++ b/libmultipath/discovery.h
@@ -42,7 +42,7 @@ int alloc_path_with_pathinfo (struct config *conf, struct udev_device *udevice,
int store_pathinfo (vector pathvec, struct config *conf,
struct udev_device *udevice, int flag,
struct path **pp_ptr);
-int sysfs_set_scsi_tmo (struct multipath *mpp, unsigned int checkint);
+int sysfs_set_scsi_tmo (struct config *conf, struct multipath *mpp);
int sysfs_get_timeout(const struct path *pp, unsigned int *timeout);
int sysfs_get_host_pci_name(const struct path *pp, char *pci_name);
int sysfs_get_iscsi_ip_address(const struct path *pp, char *ip_address);
diff --git a/libmultipath/propsel.c b/libmultipath/propsel.c
index 127b3370..25326eb6 100644
--- a/libmultipath/propsel.c
+++ b/libmultipath/propsel.c
@@ -766,51 +766,51 @@ int select_minio(struct config *conf, struct multipath *mp)
return select_minio_bio(conf, mp);
}
-int select_fast_io_fail(struct config *conf, struct multipath *mp)
+int select_fast_io_fail(struct config *conf, struct path *pp)
{
const char *origin;
char buff[12];
- mp_set_ovr(fast_io_fail);
- mp_set_hwe(fast_io_fail);
- mp_set_conf(fast_io_fail);
- mp_set_default(fast_io_fail, DEFAULT_FAST_IO_FAIL);
+ pp_set_ovr(fast_io_fail);
+ pp_set_hwe(fast_io_fail);
+ pp_set_conf(fast_io_fail);
+ pp_set_default(fast_io_fail, DEFAULT_FAST_IO_FAIL);
out:
- print_undef_off_zero(buff, 12, mp->fast_io_fail);
- condlog(3, "%s: fast_io_fail_tmo = %s %s", mp->alias, buff, origin);
+ print_undef_off_zero(buff, 12, pp->fast_io_fail);
+ condlog(3, "%s: fast_io_fail_tmo = %s %s", pp->dev, buff, origin);
return 0;
}
-int select_dev_loss(struct config *conf, struct multipath *mp)
+int select_dev_loss(struct config *conf, struct path *pp)
{
const char *origin;
char buff[12];
- mp_set_ovr(dev_loss);
- mp_set_hwe(dev_loss);
- mp_set_conf(dev_loss);
- mp->dev_loss = 0;
+ pp_set_ovr(dev_loss);
+ pp_set_hwe(dev_loss);
+ pp_set_conf(dev_loss);
+ pp->dev_loss = 0;
return 0;
out:
- print_dev_loss(buff, 12, mp->dev_loss);
- condlog(3, "%s: dev_loss_tmo = %s %s", mp->alias, buff, origin);
+ print_dev_loss(buff, 12, pp->dev_loss);
+ condlog(3, "%s: dev_loss_tmo = %s %s", pp->dev, buff, origin);
return 0;
}
-int select_eh_deadline(struct config *conf, struct multipath *mp)
+int select_eh_deadline(struct config *conf, struct path *pp)
{
const char *origin;
char buff[12];
- mp_set_ovr(eh_deadline);
- mp_set_hwe(eh_deadline);
- mp_set_conf(eh_deadline);
- mp->eh_deadline = EH_DEADLINE_UNSET;
+ pp_set_ovr(eh_deadline);
+ pp_set_hwe(eh_deadline);
+ pp_set_conf(eh_deadline);
+ pp->eh_deadline = EH_DEADLINE_UNSET;
/* not changing sysfs in default cause, so don't print anything */
return 0;
out:
- print_undef_off_zero(buff, 12, mp->eh_deadline);
- condlog(3, "%s: eh_deadline = %s %s", mp->alias, buff, origin);
+ print_undef_off_zero(buff, 12, pp->eh_deadline);
+ condlog(3, "%s: eh_deadline = %s %s", pp->dev, buff, origin);
return 0;
}
diff --git a/libmultipath/propsel.h b/libmultipath/propsel.h
index 72a7e33c..152ca44c 100644
--- a/libmultipath/propsel.h
+++ b/libmultipath/propsel.h
@@ -16,9 +16,9 @@ int select_minio(struct config *conf, struct multipath *mp);
int select_mode(struct config *conf, struct multipath *mp);
int select_uid(struct config *conf, struct multipath *mp);
int select_gid(struct config *conf, struct multipath *mp);
-int select_fast_io_fail(struct config *conf, struct multipath *mp);
-int select_dev_loss(struct config *conf, struct multipath *mp);
-int select_eh_deadline(struct config *conf, struct multipath *mp);
+int select_fast_io_fail(struct config *conf, struct path *pp);
+int select_dev_loss(struct config *conf, struct path *pp);
+int select_eh_deadline(struct config *conf, struct path *pp);
int select_reservation_key(struct config *conf, struct multipath *mp);
int select_retain_hwhandler (struct config *conf, struct multipath * mp);
int select_detect_prio(struct config *conf, struct path * pp);
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index 19099bed..9f86eb69 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -230,7 +230,6 @@ alloc_multipath (void)
mpp->bestpg = 1;
mpp->mpcontext = NULL;
mpp->no_path_retry = NO_PATH_RETRY_UNDEF;
- mpp->fast_io_fail = MP_FAST_IO_FAIL_UNSET;
dm_multipath_to_gen(mpp)->ops = &dm_gen_multipath_ops;
}
return mpp;
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index ab99852f..875e726e 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -344,6 +344,9 @@ struct path {
int marginal;
int vpd_vendor_id;
int recheck_wwid;
+ int fast_io_fail;
+ unsigned int dev_loss;
+ int eh_deadline;
/* configlet pointers */
vector hwe;
struct gen_path generic_path;
@@ -371,7 +374,6 @@ struct multipath {
int minio;
int flush_on_last_del;
int attribute_flags;
- int fast_io_fail;
int retain_hwhandler;
int deferred_remove;
bool in_recovery;
@@ -389,8 +391,6 @@ struct multipath {
int needs_paths_uevent;
int ghost_delay;
int ghost_delay_tick;
- unsigned int dev_loss;
- int eh_deadline;
uid_t uid;
gid_t gid;
mode_t mode;

View File

@ -0,0 +1,84 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:39 -0500
Subject: [PATCH] libmultipath: check the overrides pctable for path variables
Paths will now also check the pctable when checking for attribtes that
exists in both the overrides section and the protocol subsection. Values
in a matching pcentry will be used in preference to values in the
overrides hwentry.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/propsel.c | 29 ++++++++++++++++++++++++++---
1 file changed, 26 insertions(+), 3 deletions(-)
diff --git a/libmultipath/propsel.c b/libmultipath/propsel.c
index 25326eb6..209c1d67 100644
--- a/libmultipath/propsel.c
+++ b/libmultipath/propsel.c
@@ -79,6 +79,8 @@ static const char conf_origin[] =
"(setting: multipath.conf defaults/devices section)";
static const char overrides_origin[] =
"(setting: multipath.conf overrides section)";
+static const char overrides_pce_origin[] =
+ "(setting: multipath.conf overrides protocol section)";
static const char cmdline_origin[] =
"(setting: multipath command line [-p] flag)";
static const char autodetect_origin[] =
@@ -144,6 +146,27 @@ do { \
} \
} while (0)
+#define pp_set_ovr_pce(var) \
+do { \
+ struct pcentry *_pce; \
+ int _i; \
+ \
+ if (conf->overrides) { \
+ vector_foreach_slot(conf->overrides->pctable, _pce, _i) { \
+ if (_pce->type == (int)bus_protocol_id(pp) && _pce->var) { \
+ pp->var = _pce->var; \
+ origin = overrides_pce_origin; \
+ goto out; \
+ } \
+ } \
+ if (conf->overrides->var) { \
+ pp->var = conf->overrides->var; \
+ origin = overrides_origin; \
+ goto out; \
+ } \
+ } \
+} while (0)
+
int select_mode(struct config *conf, struct multipath *mp)
{
const char *origin;
@@ -771,7 +794,7 @@ int select_fast_io_fail(struct config *conf, struct path *pp)
const char *origin;
char buff[12];
- pp_set_ovr(fast_io_fail);
+ pp_set_ovr_pce(fast_io_fail);
pp_set_hwe(fast_io_fail);
pp_set_conf(fast_io_fail);
pp_set_default(fast_io_fail, DEFAULT_FAST_IO_FAIL);
@@ -786,7 +809,7 @@ int select_dev_loss(struct config *conf, struct path *pp)
const char *origin;
char buff[12];
- pp_set_ovr(dev_loss);
+ pp_set_ovr_pce(dev_loss);
pp_set_hwe(dev_loss);
pp_set_conf(dev_loss);
pp->dev_loss = 0;
@@ -802,7 +825,7 @@ int select_eh_deadline(struct config *conf, struct path *pp)
const char *origin;
char buff[12];
- pp_set_ovr(eh_deadline);
+ pp_set_ovr_pce(eh_deadline);
pp_set_hwe(eh_deadline);
pp_set_conf(eh_deadline);
pp->eh_deadline = EH_DEADLINE_UNSET;

View File

@ -0,0 +1,33 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:40 -0500
Subject: [PATCH] libmultipath: fix eh_deadline documentation
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipath/multipath.conf.5 | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index 0c04c7e4..b14fd215 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -1602,6 +1602,8 @@ section:
.TP
.B dev_loss_tmo
.TP
+.B eh_deadline
+.TP
.B flush_on_last_del
.TP
.B user_friendly_names
@@ -1688,6 +1690,8 @@ the values are taken from the \fIdevices\fR or \fIdefaults\fR sections:
.TP
.B dev_loss_tmo
.TP
+.B eh_deadline
+.TP
.B user_friendly_names
.TP
.B retain_attached_hw_handler

View File

@ -0,0 +1,54 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 13 Apr 2022 23:27:41 -0500
Subject: [PATCH] libmultipath: Add documentation for the protocol subsection
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipath/multipath.conf.5 | 32 ++++++++++++++++++++++++++++++++
1 file changed, 32 insertions(+)
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index b14fd215..abbc89af 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -1730,6 +1730,38 @@ the values are taken from the \fIdevices\fR or \fIdefaults\fR sections:
.RE
.PD
.LP
+The overrides section also recognizes the optional \fIprotocol\fR subsection,
+and can contain multiple protocol subsections. Path devices are matched against
+the protocol subsection using the mandatory \fItype\fR attribute. Attributes
+in a matching protocol subsection take precedence over attributes in the rest
+of the overrides section. If there are multiple matching protocol subsections,
+later entries take precedence.
+.TP
+.B protocol subsection
+The protocol subsection recognizes the following mandatory attribute:
+.RS
+.TP
+.B type
+The protocol string of the path device. The possible values are \fIscsi:fcp\fR,
+\fIscsi:spi\fR, \fIscsi:ssa\fR, \fIscsi:sbp\fR, \fIscsi:srp\fR,
+\fIscsi:iscsi\fR, \fIscsi:sas\fR, \fIscsi:adt\fR, \fIscsi:ata\fR,
+\fIscsi:unspec\fR, \fIccw\fR, \fIcciss\fR, \fInvme\fR, and \fIundef\fR. This is
+\fBnot\fR a regular expression. the path device protcol string must match
+exactly. The protocol that a path is using can be viewed by running
+\fBmultipathd show paths format "%d %P"\fR
+.LP
+The following attributes are optional; if not set, the default values are taken
+from the \fIoverrides\fR, \fIdevices\fR, or \fIdefaults\fR section:
+.sp 1
+.PD .1v
+.RS
+.TP
+.B fast_io_fail_tmo
+.TP
+.B dev_loss_tmo
+.TP
+.B eh_deadline
+.PD
.
.
.\" ----------------------------------------------------------------------------

View File

@ -0,0 +1,47 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Thu, 14 Apr 2022 16:11:10 -0500
Subject: [PATCH] libmultipath: use symbolic value for invalid pcentry
Suggested-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/config.c | 4 ++--
libmultipath/config.h | 1 +
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/libmultipath/config.c b/libmultipath/config.c
index 88975323..bcd8b541 100644
--- a/libmultipath/config.c
+++ b/libmultipath/config.c
@@ -319,7 +319,7 @@ alloc_pce (void)
{
struct pcentry *pce = (struct pcentry *)
calloc(1, sizeof(struct pcentry));
- pce->type = -1;
+ pce->type = PCE_INVALID;
return pce;
}
@@ -578,7 +578,7 @@ validate_pctable(struct hwentry *ovr, int idx, const char *table_desc)
return;
vector_foreach_slot_after(ovr->pctable, pce, idx) {
- if (pce->type < 0) {
+ if (pce->type == PCE_INVALID) {
condlog(0, "protocol section in %s missing type",
table_desc);
vector_del_slot(ovr->pctable, idx--);
diff --git a/libmultipath/config.h b/libmultipath/config.h
index 143116b3..477f8bfa 100644
--- a/libmultipath/config.h
+++ b/libmultipath/config.h
@@ -41,6 +41,7 @@ enum force_reload_types {
FORCE_RELOAD_WEAK,
};
+#define PCE_INVALID -1
struct pcentry {
int type;
int fast_io_fail;

File diff suppressed because it is too large Load Diff

View File

@ -0,0 +1,73 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 11 Feb 2022 17:23:39 -0600
Subject: [PATCH] multipathd: disallow changing to/from fpin marginal paths on
reconfig
Setting marginal_pathgroups to fpin causes two new threads to be created
when multipathd starts. Turning it on after multipathd starts up won't
cause the theads to start, and turing it off won't keep the threads from
working. So disallow changing marginal_pathgroups to/from "fpin" on
reconfigure.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
multipath/multipath.conf.5 | 13 ++++++++-----
multipathd/main.c | 9 +++++++++
2 files changed, 17 insertions(+), 5 deletions(-)
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index 805b7a5e..8e418372 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -1077,15 +1077,18 @@ have been tried first. This prevents the possibility of IO errors occuring
while marginal paths are still usable. After the path has been monitored
for the configured time, and is declared healthy, it will be returned to its
normal pathgroup.
-However if this option is set to \fIfpin\fR multipathd will receive fpin
+If this option is set to \fIfpin\fR, multipathd will receive fpin
notifications, set path states to "marginal" accordingly, and regroup paths
-as described for "marginal_pathgroups yes". This option can't be used in combination
-with other options for "Shaky path detection" (see below).If it is set to fpin,
-marginal_path_xyz and san_path_err_xyz parameters are implicitly set to 0.
+as described for \fIon\fR. This option can't be used in combination
+with other options for "Shaky path detection" (see below). \fBNote:\fR If this
+is set to \fIfpin\fR, the \fImarginal_path_*\fR and \fIsan_path_err_*\fR
+options are implicitly set to \fIno\fP. Also, this option cannot be switched
+either to or from \fIfpin\fR on a multipathd reconfigure. multipathd must be
+restarted for the change to take effect.
See "Shaky paths detection" below for more information.
.RS
.TP
-The default is: \fBno\fR
+The default is: \fBoff\fR
.RE
.
.
diff --git a/multipathd/main.c b/multipathd/main.c
index 4cf5bc41..a6ffbe32 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -2675,6 +2675,7 @@ int
reconfigure (struct vectors * vecs)
{
struct config * old, *conf;
+ int old_marginal_pathgroups;
conf = load_config(DEFAULT_CONFIGFILE);
if (!conf)
@@ -2702,6 +2703,14 @@ reconfigure (struct vectors * vecs)
uxsock_timeout = conf->uxsock_timeout;
old = rcu_dereference(multipath_conf);
+ old_marginal_pathgroups = old->marginal_pathgroups;
+ if ((old_marginal_pathgroups == MARGINAL_PATHGROUP_FPIN) !=
+ (conf->marginal_pathgroups == MARGINAL_PATHGROUP_FPIN)) {
+ condlog(1, "multipathd must be restarted to turn %s fpin marginal paths",
+ (old_marginal_pathgroups == MARGINAL_PATHGROUP_FPIN)?
+ "off" : "on");
+ conf->marginal_pathgroups = old_marginal_pathgroups;
+ }
conf->sequence_nr = old->sequence_nr + 1;
rcu_assign_pointer(multipath_conf, conf);
call_rcu(&old->rcu, rcu_free_config);

View File

@ -0,0 +1,30 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 10 May 2022 14:17:22 -0500
Subject: [PATCH] libmultipath: unset detect_checker for clariion / Unity
arrays
Dell EMC would like to always use the emc_clariion checker. Currently
detect_checker will switch the checker to TUR for Unity arrays.
This can cause problems on some setups with replicated Unity LUNs,
which are handled correctly the the emc_checker, but not the TUR
checker.
Cc: vincent.chen1@dell.com
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/hwtable.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libmultipath/hwtable.c b/libmultipath/hwtable.c
index d6325864..2bb274c5 100644
--- a/libmultipath/hwtable.c
+++ b/libmultipath/hwtable.c
@@ -343,6 +343,7 @@ static struct hwentry default_hw[] = {
.no_path_retry = (300 / DEFAULT_CHECKINT),
.checker_name = EMC_CLARIION,
.prio_name = PRIO_EMC,
+ .detect_checker = DETECT_CHECKER_OFF,
},
{
/* Invista / VPLEX */

View File

@ -0,0 +1,26 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Bastian Germann <bage@debian.org>
Date: Thu, 14 Oct 2021 00:34:33 +0200
Subject: [PATCH] multipathd: Add missing ctype include
In uxclnt.c, there are isspace calls. Add an explicit include.
Signed-off-by: Bastian Germann <bage@debian.org>
Reviewed-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
multipathd/uxclnt.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/multipathd/uxclnt.c b/multipathd/uxclnt.c
index a76f8e29..f16a7309 100644
--- a/multipathd/uxclnt.c
+++ b/multipathd/uxclnt.c
@@ -8,6 +8,7 @@
#include <stdlib.h>
#include <unistd.h>
#include <stdarg.h>
+#include <ctype.h>
#include <fcntl.h>
#include <errno.h>
#include <sys/ioctl.h>

View File

@ -0,0 +1,102 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Fri, 12 Aug 2022 18:58:15 +0200
Subject: [PATCH] multipathd: replace libreadline with libedit
Linking multipathd with libreadline may cause a license conflict,
because libreadline is licensed under GPL-3.0-or-later, and
libmultipath contains several files under GPL-2.0.
See:
https://github.com/opensvc/multipath-tools/issues/36
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=979095
https://www.gnu.org/licenses/gpl-faq.html#AllCompatibility
Replace the readline functionality with libedit, which comes under
a BSD license. The readline library can still be enabled (e.g. for
binaries not intended to be distributed) by running
"make READLINE=libreadline".
Signed-off-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
Makefile.inc | 5 +++++
multipathd/Makefile | 11 ++++++++++-
multipathd/cli.c | 5 +++++
multipathd/uxclnt.c | 6 ++++++
4 files changed, 26 insertions(+), 1 deletion(-)
diff --git a/Makefile.inc b/Makefile.inc
index 25c16f4e..d471f045 100644
--- a/Makefile.inc
+++ b/Makefile.inc
@@ -14,6 +14,11 @@
#
# Uncomment to disable dmevents polling support
# ENABLE_DMEVENTS_POLL = 0
+#
+# Readline library to use, libedit or libreadline
+# Caution: Using libreadline may make the multipathd binary undistributable,
+# see https://github.com/opensvc/multipath-tools/issues/36
+READLINE = libedit
ifeq ($(TOPDIR),)
TOPDIR = ..
diff --git a/multipathd/Makefile b/multipathd/Makefile
index 835edd93..4874ec3a 100644
--- a/multipathd/Makefile
+++ b/multipathd/Makefile
@@ -15,7 +15,16 @@ CFLAGS += $(BIN_CFLAGS) -I$(multipathdir) -I$(mpathpersistdir) \
LDFLAGS += $(BIN_LDFLAGS)
LIBDEPS += -L$(multipathdir) -lmultipath -L$(mpathpersistdir) -lmpathpersist \
-L$(mpathcmddir) -lmpathcmd -ludev -ldl -lurcu -lpthread \
- -ldevmapper -lreadline
+ -ldevmapper
+
+ifeq ($(READLINE),libedit)
+CFLAGS += -DUSE_LIBEDIT
+LIBDEPS += -ledit
+endif
+ifeq ($(READLINE),libreadline)
+CFLAGS += -DUSE_LIBREADLINE
+LIBDEPS += -lreadline
+endif
ifdef SYSTEMD
CFLAGS += -DUSE_SYSTEMD=$(SYSTEMD)
diff --git a/multipathd/cli.c b/multipathd/cli.c
index bdc9fb10..85d73dfb 100644
--- a/multipathd/cli.c
+++ b/multipathd/cli.c
@@ -11,7 +11,12 @@
#include "parser.h"
#include "util.h"
#include "version.h"
+#ifdef USE_LIBEDIT
+#include <editline/readline.h>
+#endif
+#ifdef USE_LIBREADLINE
#include <readline/readline.h>
+#endif
#include "mpath_cmd.h"
#include "cli.h"
diff --git a/multipathd/uxclnt.c b/multipathd/uxclnt.c
index f16a7309..2c17d8fc 100644
--- a/multipathd/uxclnt.c
+++ b/multipathd/uxclnt.c
@@ -16,8 +16,14 @@
#include <sys/socket.h>
#include <sys/un.h>
#include <poll.h>
+
+#ifdef USE_LIBEDIT
+#include <editline/readline.h>
+#endif
+#ifdef USE_LIBREADLINE
#include <readline/readline.h>
#include <readline/history.h>
+#endif
#include "mpath_cmd.h"
#include "uxsock.h"

View File

@ -0,0 +1,35 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 5 Aug 2022 18:16:03 -0500
Subject: [PATCH] multipath: fix systemd timers in the initramfs
The systemd timers created for "find_multipaths smart" conflict with
shutdown.target, but not with initrd-cleanup.service. This can make
these timers trigger after the inirtd has started shutting down,
restarting multipathd (which then stops initrd-cleanup.service, since it
conflicts). To avoid this, make sure the timers and the unit they
trigger conflict with inird-cleanup.service. Also don't make them start
multipathd. "multipath -u" will not return "maybe" if multipathd isn't
running or set to run, and since we no longer wait for udev-settle,
multipathd starts up pretty quickly, so it shouldn't be a problem to
not trigger it here.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipath/multipath.rules | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/multipath/multipath.rules b/multipath/multipath.rules
index 0486bf70..68c30644 100644
--- a/multipath/multipath.rules
+++ b/multipath/multipath.rules
@@ -72,7 +72,7 @@ ENV{.SAVED_FM_WAIT_UNTIL}=="?*", GOTO="pretend_mpath"
#
# We must trigger an "add" event because LVM2 will only act on those.
-RUN+="/usr/bin/systemd-run --unit=cancel-multipath-wait-$kernel --description 'cancel waiting for multipath siblings of $kernel' --no-block --timer-property DefaultDependencies=no --timer-property Conflicts=shutdown.target --timer-property Before=shutdown.target --timer-property AccuracySec=500ms --property DefaultDependencies=no --property Conflicts=shutdown.target --property Before=shutdown.target --property Wants=multipathd.service --property After=multipathd.service --on-active=$env{FIND_MULTIPATHS_WAIT_UNTIL} /usr/bin/udevadm trigger --action=add $sys$devpath"
+RUN+="/usr/bin/systemd-run --unit=cancel-multipath-wait-$kernel --description 'cancel waiting for multipath siblings of $kernel' --no-block --timer-property DefaultDependencies=no --timer-property Conflicts=shutdown.target --timer-property Before=shutdown.target --timer-property Conflicts=initrd-cleanup.service --timer-property Before=initrd-cleanup.service --timer-property AccuracySec=500ms --property DefaultDependencies=no --property Conflicts=shutdown.target --property Before=shutdown.target --property Conflicts=initrd-cleanup.service --property Before=initrd-cleanup.service --on-active=$env{FIND_MULTIPATHS_WAIT_UNTIL} /usr/bin/udevadm trigger --action=add $sys$devpath"
LABEL="pretend_mpath"
ENV{DM_MULTIPATH_DEVICE_PATH}="1"

View File

@ -0,0 +1,90 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 9 Aug 2022 16:46:26 -0500
Subject: [PATCH] multipathd: factor out the code to flush a map with no paths
The code to flush a multipath device because all of its paths have
been removed will be used by another caller, so factor it out of
ev_remove_path().
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipathd/main.c | 56 ++++++++++++++++++++++++-----------------------
1 file changed, 29 insertions(+), 27 deletions(-)
diff --git a/multipathd/main.c b/multipathd/main.c
index a6ffbe32..9b1098f6 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -487,6 +487,30 @@ int update_multipath (struct vectors *vecs, char *mapname, int reset)
return 0;
}
+static bool
+flush_map_nopaths(struct multipath *mpp, struct vectors *vecs) {
+ char alias[WWID_SIZE];
+
+ /*
+ * flush_map will fail if the device is open
+ */
+ strlcpy(alias, mpp->alias, WWID_SIZE);
+ if (mpp->flush_on_last_del == FLUSH_ENABLED) {
+ condlog(2, "%s Last path deleted, disabling queueing",
+ mpp->alias);
+ mpp->retry_tick = 0;
+ mpp->no_path_retry = NO_PATH_RETRY_FAIL;
+ mpp->disable_queueing = 1;
+ mpp->stat_map_failures++;
+ dm_queue_if_no_path(mpp->alias, 0);
+ }
+ if (!flush_map(mpp, vecs, 1)) {
+ condlog(2, "%s: removed map after removing all paths", alias);
+ return true;
+ }
+ return false;
+}
+
static int
update_map (struct multipath *mpp, struct vectors *vecs, int new_map)
{
@@ -1185,34 +1209,12 @@ ev_remove_path (struct path *pp, struct vectors * vecs, int need_do_map)
vector_del_slot(mpp->paths, i);
/*
- * remove the map IF removing the last path
+ * remove the map IF removing the last path. If
+ * flush_map_nopaths succeeds, the path has been removed.
*/
- if (VECTOR_SIZE(mpp->paths) == 0) {
- char alias[WWID_SIZE];
-
- /*
- * flush_map will fail if the device is open
- */
- strlcpy(alias, mpp->alias, WWID_SIZE);
- if (mpp->flush_on_last_del == FLUSH_ENABLED) {
- condlog(2, "%s Last path deleted, disabling queueing", mpp->alias);
- mpp->retry_tick = 0;
- mpp->no_path_retry = NO_PATH_RETRY_FAIL;
- mpp->disable_queueing = 1;
- mpp->stat_map_failures++;
- dm_queue_if_no_path(mpp->alias, 0);
- }
- if (!flush_map(mpp, vecs, 1)) {
- condlog(2, "%s: removed map after"
- " removing all paths",
- alias);
- retval = 0;
- goto out;
- }
- /*
- * Not an error, continue
- */
- }
+ if (VECTOR_SIZE(mpp->paths) == 0 &&
+ flush_map_nopaths(mpp, vecs))
+ goto out;
if (mpp->hwe == NULL)
extract_hwe_from_path(mpp);

View File

@ -0,0 +1,47 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 9 Aug 2022 16:46:27 -0500
Subject: [PATCH] libmultipath: return success if we raced to remove a map and
lost
_dm_flush_map() was returning failure if it failed to remove a map,
even if that was because the map had already been removed. Return
success in this case.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/devmapper.c | 4 ++++
multipathd/main.c | 4 ----
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/libmultipath/devmapper.c b/libmultipath/devmapper.c
index e3c3c119..ee81acc3 100644
--- a/libmultipath/devmapper.c
+++ b/libmultipath/devmapper.c
@@ -916,6 +916,10 @@ int _dm_flush_map (const char * mapname, int need_sync, int deferred_remove,
}
condlog(4, "multipath map %s removed", mapname);
return 0;
+ } else if (dm_is_mpath(mapname) != 1) {
+ condlog(4, "multipath map %s removed externally",
+ mapname);
+ return 0; /*we raced with someone else removing it */
} else {
condlog(2, "failed to remove multipath map %s",
mapname);
diff --git a/multipathd/main.c b/multipathd/main.c
index 9b1098f6..9eafd5b7 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -673,10 +673,6 @@ flush_map(struct multipath * mpp, struct vectors * vecs, int nopaths)
* the spurious uevent we may generate with the dm_flush_map call below
*/
if (r) {
- /*
- * May not really be an error -- if the map was already flushed
- * from the device mapper by dmsetup(8) for instance.
- */
if (r == 1)
condlog(0, "%s: can't flush", mpp->alias);
else {

View File

@ -0,0 +1,36 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 9 Aug 2022 16:46:28 -0500
Subject: [PATCH] multipathd: Handle losing all path in update_map
Its possible that when a multipath device is being updated, it will end
up that all the paths for it are gone. This can happen if paths are
added and then removed again before multipathd processes the uevent for
the newly created multipath device. In this case multipathd wasn't
taking the proper action for the case where all the paths had been
removed. If flush_on_last_del was set, multipathd wasn't disabling
flushing and if deferred_remove was set, it wasn't doing a deferred
remove. Multipathd should call flush_map_nopaths(), just like
ev_remove_path() does when the last path is removed.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipathd/main.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/multipathd/main.c b/multipathd/main.c
index 9eafd5b7..870ae7d8 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -526,6 +526,10 @@ retry:
goto fail;
}
verify_paths(mpp, vecs);
+ if (VECTOR_SIZE(mpp->paths) == 0 &&
+ flush_map_nopaths(mpp, vecs))
+ return 1;
+
mpp->action = ACT_RELOAD;
if (setup_map(mpp, params, PARAMS_SIZE, vecs)) {

View File

@ -0,0 +1,153 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Thu, 8 Sep 2022 11:54:49 -0500
Subject: [PATCH] multipathd: ignore duplicated multipathd command keys
multipath adds rather than or-s the values of command keys. Fix this.
Also, return an invalid fingerprint if a key is used more than once.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
multipathd/cli.c | 8 ++--
multipathd/main.c | 106 +++++++++++++++++++++++-----------------------
2 files changed, 58 insertions(+), 56 deletions(-)
diff --git a/multipathd/cli.c b/multipathd/cli.c
index 85d73dfb..45e80197 100644
--- a/multipathd/cli.c
+++ b/multipathd/cli.c
@@ -341,9 +341,11 @@ fingerprint(vector vec)
if (!vec)
return 0;
- vector_foreach_slot(vec, kw, i)
- fp += kw->code;
-
+ vector_foreach_slot(vec, kw, i) {
+ if (fp & kw->code)
+ return (uint64_t)-1;
+ fp |= kw->code;
+ }
return fp;
}
diff --git a/multipathd/main.c b/multipathd/main.c
index 870ae7d8..cd68a9d2 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -1628,62 +1628,62 @@ uxlsnrloop (void * ap)
/* Tell main thread that thread has started */
post_config_state(DAEMON_CONFIGURE);
- set_handler_callback(LIST+PATHS, cli_list_paths);
- set_handler_callback(LIST+PATHS+FMT, cli_list_paths_fmt);
- set_handler_callback(LIST+PATHS+RAW+FMT, cli_list_paths_raw);
- set_handler_callback(LIST+PATH, cli_list_path);
- set_handler_callback(LIST+MAPS, cli_list_maps);
- set_handler_callback(LIST+STATUS, cli_list_status);
- set_unlocked_handler_callback(LIST+DAEMON, cli_list_daemon);
- set_handler_callback(LIST+MAPS+STATUS, cli_list_maps_status);
- set_handler_callback(LIST+MAPS+STATS, cli_list_maps_stats);
- set_handler_callback(LIST+MAPS+FMT, cli_list_maps_fmt);
- set_handler_callback(LIST+MAPS+RAW+FMT, cli_list_maps_raw);
- set_handler_callback(LIST+MAPS+TOPOLOGY, cli_list_maps_topology);
- set_handler_callback(LIST+TOPOLOGY, cli_list_maps_topology);
- set_handler_callback(LIST+MAPS+JSON, cli_list_maps_json);
- set_handler_callback(LIST+MAP+TOPOLOGY, cli_list_map_topology);
- set_handler_callback(LIST+MAP+FMT, cli_list_map_fmt);
- set_handler_callback(LIST+MAP+RAW+FMT, cli_list_map_fmt);
- set_handler_callback(LIST+MAP+JSON, cli_list_map_json);
- set_handler_callback(LIST+CONFIG+LOCAL, cli_list_config_local);
- set_handler_callback(LIST+CONFIG, cli_list_config);
- set_handler_callback(LIST+BLACKLIST, cli_list_blacklist);
- set_handler_callback(LIST+DEVICES, cli_list_devices);
- set_handler_callback(LIST+WILDCARDS, cli_list_wildcards);
- set_handler_callback(RESET+MAPS+STATS, cli_reset_maps_stats);
- set_handler_callback(RESET+MAP+STATS, cli_reset_map_stats);
- set_handler_callback(ADD+PATH, cli_add_path);
- set_handler_callback(DEL+PATH, cli_del_path);
- set_handler_callback(ADD+MAP, cli_add_map);
- set_handler_callback(DEL+MAP, cli_del_map);
- set_handler_callback(DEL+MAPS, cli_del_maps);
- set_handler_callback(SWITCH+MAP+GROUP, cli_switch_group);
+ set_handler_callback(LIST|PATHS, cli_list_paths);
+ set_handler_callback(LIST|PATHS|FMT, cli_list_paths_fmt);
+ set_handler_callback(LIST|PATHS|RAW|FMT, cli_list_paths_raw);
+ set_handler_callback(LIST|PATH, cli_list_path);
+ set_handler_callback(LIST|MAPS, cli_list_maps);
+ set_handler_callback(LIST|STATUS, cli_list_status);
+ set_unlocked_handler_callback(LIST|DAEMON, cli_list_daemon);
+ set_handler_callback(LIST|MAPS|STATUS, cli_list_maps_status);
+ set_handler_callback(LIST|MAPS|STATS, cli_list_maps_stats);
+ set_handler_callback(LIST|MAPS|FMT, cli_list_maps_fmt);
+ set_handler_callback(LIST|MAPS|RAW|FMT, cli_list_maps_raw);
+ set_handler_callback(LIST|MAPS|TOPOLOGY, cli_list_maps_topology);
+ set_handler_callback(LIST|TOPOLOGY, cli_list_maps_topology);
+ set_handler_callback(LIST|MAPS|JSON, cli_list_maps_json);
+ set_handler_callback(LIST|MAP|TOPOLOGY, cli_list_map_topology);
+ set_handler_callback(LIST|MAP|FMT, cli_list_map_fmt);
+ set_handler_callback(LIST|MAP|RAW|FMT, cli_list_map_fmt);
+ set_handler_callback(LIST|MAP|JSON, cli_list_map_json);
+ set_handler_callback(LIST|CONFIG|LOCAL, cli_list_config_local);
+ set_handler_callback(LIST|CONFIG, cli_list_config);
+ set_handler_callback(LIST|BLACKLIST, cli_list_blacklist);
+ set_handler_callback(LIST|DEVICES, cli_list_devices);
+ set_handler_callback(LIST|WILDCARDS, cli_list_wildcards);
+ set_handler_callback(RESET|MAPS|STATS, cli_reset_maps_stats);
+ set_handler_callback(RESET|MAP|STATS, cli_reset_map_stats);
+ set_handler_callback(ADD|PATH, cli_add_path);
+ set_handler_callback(DEL|PATH, cli_del_path);
+ set_handler_callback(ADD|MAP, cli_add_map);
+ set_handler_callback(DEL|MAP, cli_del_map);
+ set_handler_callback(DEL|MAPS, cli_del_maps);
+ set_handler_callback(SWITCH|MAP|GROUP, cli_switch_group);
set_unlocked_handler_callback(RECONFIGURE, cli_reconfigure);
- set_handler_callback(SUSPEND+MAP, cli_suspend);
- set_handler_callback(RESUME+MAP, cli_resume);
- set_handler_callback(RESIZE+MAP, cli_resize);
- set_handler_callback(RELOAD+MAP, cli_reload);
- set_handler_callback(RESET+MAP, cli_reassign);
- set_handler_callback(REINSTATE+PATH, cli_reinstate);
- set_handler_callback(FAIL+PATH, cli_fail);
- set_handler_callback(DISABLEQ+MAP, cli_disable_queueing);
- set_handler_callback(RESTOREQ+MAP, cli_restore_queueing);
- set_handler_callback(DISABLEQ+MAPS, cli_disable_all_queueing);
- set_handler_callback(RESTOREQ+MAPS, cli_restore_all_queueing);
+ set_handler_callback(SUSPEND|MAP, cli_suspend);
+ set_handler_callback(RESUME|MAP, cli_resume);
+ set_handler_callback(RESIZE|MAP, cli_resize);
+ set_handler_callback(RELOAD|MAP, cli_reload);
+ set_handler_callback(RESET|MAP, cli_reassign);
+ set_handler_callback(REINSTATE|PATH, cli_reinstate);
+ set_handler_callback(FAIL|PATH, cli_fail);
+ set_handler_callback(DISABLEQ|MAP, cli_disable_queueing);
+ set_handler_callback(RESTOREQ|MAP, cli_restore_queueing);
+ set_handler_callback(DISABLEQ|MAPS, cli_disable_all_queueing);
+ set_handler_callback(RESTOREQ|MAPS, cli_restore_all_queueing);
set_unlocked_handler_callback(QUIT, cli_quit);
set_unlocked_handler_callback(SHUTDOWN, cli_shutdown);
- set_handler_callback(GETPRSTATUS+MAP, cli_getprstatus);
- set_handler_callback(SETPRSTATUS+MAP, cli_setprstatus);
- set_handler_callback(UNSETPRSTATUS+MAP, cli_unsetprstatus);
- set_handler_callback(FORCEQ+DAEMON, cli_force_no_daemon_q);
- set_handler_callback(RESTOREQ+DAEMON, cli_restore_no_daemon_q);
- set_handler_callback(GETPRKEY+MAP, cli_getprkey);
- set_handler_callback(SETPRKEY+MAP+KEY, cli_setprkey);
- set_handler_callback(UNSETPRKEY+MAP, cli_unsetprkey);
- set_handler_callback(SETMARGINAL+PATH, cli_set_marginal);
- set_handler_callback(UNSETMARGINAL+PATH, cli_unset_marginal);
- set_handler_callback(UNSETMARGINAL+MAP, cli_unset_all_marginal);
+ set_handler_callback(GETPRSTATUS|MAP, cli_getprstatus);
+ set_handler_callback(SETPRSTATUS|MAP, cli_setprstatus);
+ set_handler_callback(UNSETPRSTATUS|MAP, cli_unsetprstatus);
+ set_handler_callback(FORCEQ|DAEMON, cli_force_no_daemon_q);
+ set_handler_callback(RESTOREQ|DAEMON, cli_restore_no_daemon_q);
+ set_handler_callback(GETPRKEY|MAP, cli_getprkey);
+ set_handler_callback(SETPRKEY|MAP|KEY, cli_setprkey);
+ set_handler_callback(UNSETPRKEY|MAP, cli_unsetprkey);
+ set_handler_callback(SETMARGINAL|PATH, cli_set_marginal);
+ set_handler_callback(UNSETMARGINAL|PATH, cli_unset_marginal);
+ set_handler_callback(UNSETMARGINAL|MAP, cli_unset_all_marginal);
umask(077);
uxsock_listen(&uxsock_trigger, ux_sock, ap);

View File

@ -0,0 +1,141 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Thu, 1 Sep 2022 19:21:30 +0200
Subject: [PATCH] multipath-tools: use /run instead of /dev/shm
/dev/shm may have unsafe permissions. Use /run instead.
Use systemd's tmpfiles.d mechanism to create /run/multipath
early during boot.
For backward compatibilty, make the runtime directory configurable
via the "runtimedir" make variable.
Signed-off-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
.gitignore | 2 ++
Makefile.inc | 4 +++-
libmultipath/defaults.h | 2 +-
multipath/Makefile | 9 +++++++--
multipath/{multipath.rules => multipath.rules.in} | 4 ++--
multipath/tmpfiles.conf.in | 1 +
6 files changed, 16 insertions(+), 6 deletions(-)
rename multipath/{multipath.rules => multipath.rules.in} (96%)
create mode 100644 multipath/tmpfiles.conf.in
diff --git a/.gitignore b/.gitignore
index 9926756b..f90b0350 100644
--- a/.gitignore
+++ b/.gitignore
@@ -8,6 +8,8 @@
*.d
kpartx/kpartx
multipath/multipath
+multipath/multipath.rules
+multipath/tmpfiles.conf
multipathd/multipathd
mpathpersist/mpathpersist
.nfs*
diff --git a/Makefile.inc b/Makefile.inc
index d471f045..287f0e7b 100644
--- a/Makefile.inc
+++ b/Makefile.inc
@@ -60,6 +60,7 @@ exec_prefix = $(prefix)
usr_prefix = $(prefix)
bindir = $(exec_prefix)/usr/sbin
libudevdir = $(prefix)/$(SYSTEMDPATH)/udev
+tmpfilesdir = $(prefix)/$(SYSTEMDPATH)/tmpfiles.d
udevrulesdir = $(libudevdir)/rules.d
multipathdir = $(TOPDIR)/libmultipath
man8dir = $(prefix)/usr/share/man/man8
@@ -76,6 +77,7 @@ libdmmpdir = $(TOPDIR)/libdmmp
nvmedir = $(TOPDIR)/libmultipath/nvme
includedir = $(prefix)/usr/include
pkgconfdir = $(usrlibdir)/pkgconfig
+runtimedir = /$(RUN)
GZIP = gzip -9 -c
RM = rm -f
@@ -117,7 +119,7 @@ OPTFLAGS += -Werror -Wextra -Wstrict-prototypes -Wformat=2 \
-Werror=cast-qual $(ERROR_DISCARDED_QUALIFIERS) \
--param=ssp-buffer-size=4
-CFLAGS := $(OPTFLAGS) -DBIN_DIR=\"$(bindir)\" -DLIB_STRING=\"${LIB}\" -DRUN_DIR=\"${RUN}\" \
+CFLAGS := $(OPTFLAGS) -DBIN_DIR=\"$(bindir)\" -DLIB_STRING=\"${LIB}\" -DRUN_DIR=\"${RUN}\" -DRUNTIME_DIR=\"$(runtimedir)\" \
-MMD -MP $(CFLAGS)
BIN_CFLAGS = -fPIE -DPIE
LIB_CFLAGS = -fPIC
diff --git a/libmultipath/defaults.h b/libmultipath/defaults.h
index f1cb000d..dcd9f5a7 100644
--- a/libmultipath/defaults.h
+++ b/libmultipath/defaults.h
@@ -65,7 +65,7 @@
#define DEFAULT_WWIDS_FILE "/etc/multipath/wwids"
#define DEFAULT_PRKEYS_FILE "/etc/multipath/prkeys"
#define DEFAULT_CONFIG_DIR "/etc/multipath/conf.d"
-#define MULTIPATH_SHM_BASE "/dev/shm/multipath/"
+#define MULTIPATH_SHM_BASE RUNTIME_DIR "/multipath/"
static inline char *set_default(char *str)
diff --git a/multipath/Makefile b/multipath/Makefile
index e720c7f6..f3d98012 100644
--- a/multipath/Makefile
+++ b/multipath/Makefile
@@ -12,7 +12,7 @@ EXEC = multipath
OBJS = main.o
-all: $(EXEC)
+all: $(EXEC) multipath.rules tmpfiles.conf
$(EXEC): $(OBJS) $(multipathdir)/libmultipath.so $(mpathcmddir)/libmpathcmd.so
$(CC) $(CFLAGS) $(OBJS) -o $(EXEC) $(LDFLAGS) $(LIBDEPS)
@@ -27,6 +27,8 @@ install:
$(INSTALL_PROGRAM) -d $(DESTDIR)$(udevrulesdir)
$(INSTALL_PROGRAM) -m 644 11-dm-mpath.rules $(DESTDIR)$(udevrulesdir)
$(INSTALL_PROGRAM) -m 644 $(EXEC).rules $(DESTDIR)$(libudevdir)/rules.d/62-multipath.rules
+ $(INSTALL_PROGRAM) -d $(DESTDIR)$(tmpfilesdir)
+ $(INSTALL_PROGRAM) -m 644 tmpfiles.conf $(DESTDIR)$(tmpfilesdir)/multipath.conf
$(INSTALL_PROGRAM) -d $(DESTDIR)$(man8dir)
$(INSTALL_PROGRAM) -m 644 $(EXEC).8.gz $(DESTDIR)$(man8dir)
$(INSTALL_PROGRAM) -d $(DESTDIR)$(man5dir)
@@ -43,9 +45,12 @@ uninstall:
$(RM) $(DESTDIR)$(man8dir)/mpathconf.8.gz
clean: dep_clean
- $(RM) core *.o $(EXEC) *.gz
+ $(RM) core *.o $(EXEC) *.gz multipath.rules tmpfiles.conf
include $(wildcard $(OBJS:.o=.d))
dep_clean:
$(RM) $(OBJS:.o=.d)
+
+%: %.in
+ sed 's,@RUNTIME_DIR@,$(runtimedir),' $< >$@
diff --git a/multipath/multipath.rules b/multipath/multipath.rules.in
similarity index 96%
rename from multipath/multipath.rules
rename to multipath/multipath.rules.in
index 68c30644..5c4447a2 100644
--- a/multipath/multipath.rules
+++ b/multipath/multipath.rules.in
@@ -1,8 +1,8 @@
# Set DM_MULTIPATH_DEVICE_PATH if the device should be handled by multipath
SUBSYSTEM!="block", GOTO="end_mpath"
KERNEL!="sd*|dasd*|nvme*", GOTO="end_mpath"
-ACTION=="remove", TEST=="/dev/shm/multipath/find_multipaths/$major:$minor", \
- RUN+="/usr/bin/rm -f /dev/shm/multipath/find_multipaths/$major:$minor"
+ACTION=="remove", TEST=="@RUNTIME_DIR@/multipath/find_multipaths/$major:$minor", \
+ RUN+="/usr/bin/rm -f @RUNTIME_DIR@/multipath/find_multipaths/$major:$minor"
ACTION!="add|change", GOTO="end_mpath"
IMPORT{cmdline}="nompath"
diff --git a/multipath/tmpfiles.conf.in b/multipath/tmpfiles.conf.in
new file mode 100644
index 00000000..21be438a
--- /dev/null
+++ b/multipath/tmpfiles.conf.in
@@ -0,0 +1 @@
+d @RUNTIME_DIR@/multipath 0700 root root -

View File

@ -0,0 +1,47 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 19 Oct 2022 12:57:10 -0500
Subject: [PATCH] kpartx: hold device open until partitions have been created
kpartx was closing the whole device after it read the partition
information off it. This allowed a race, where the device could be
removed and another one created with the same major:minor, after kpartx
read the partition information but before it created the partition
devices.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
kpartx/kpartx.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/kpartx/kpartx.c b/kpartx/kpartx.c
index a337a07b..e62b764f 100644
--- a/kpartx/kpartx.c
+++ b/kpartx/kpartx.c
@@ -426,12 +426,7 @@ main(int argc, char **argv){
if (n >= 0)
printf("%s: %d slices\n", ptp->type, n);
#endif
-
- if (n > 0) {
- close(fd);
- fd = -1;
- }
- else
+ if (n <= 0)
continue;
switch(what) {
@@ -649,9 +644,9 @@ main(int argc, char **argv){
if (n > 0)
break;
}
+ if (fd != -1)
+ close(fd);
if (what == LIST && loopcreated) {
- if (fd != -1)
- close(fd);
if (del_loop(device)) {
if (verbose)
fprintf(stderr, "can't del loop : %s\n",

View File

@ -0,0 +1,153 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:37 -0500
Subject: [PATCH] libmultipath: cleanup remove_feature
remove_feature() didn't correctly handle feature strings that used
whitespace other than spaces, which the kernel allows. It also didn't
check if the feature string to be removed was part of a larger feature
token. Finally, it did a lot of unnecessary work. By failing if the
feature string to be removed contains leading or trailing whitespace,
the function can be significanly simplified.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/structs.c | 82 +++++++++++++++---------------------------
1 file changed, 29 insertions(+), 53 deletions(-)
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index 9f86eb69..471087e2 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -6,6 +6,7 @@
#include <unistd.h>
#include <libdevmapper.h>
#include <libudev.h>
+#include <ctype.h>
#include "checkers.h"
#include "memory.h"
@@ -633,7 +634,7 @@ int add_feature(char **f, const char *n)
int remove_feature(char **f, const char *o)
{
- int c = 0, d, l;
+ int c = 0, d;
char *e, *p, *n;
const char *q;
@@ -644,33 +645,35 @@ int remove_feature(char **f, const char *o)
if (!o || *o == '\0')
return 0;
- /* Check if not present */
- if (!strstr(*f, o))
+ d = strlen(o);
+ if (isspace(*o) || isspace(*(o + d - 1))) {
+ condlog(0, "internal error: feature \"%s\" has leading or trailing spaces", o);
+ return 1;
+ }
+
+ /* Check if present and not part of a larger feature token*/
+ p = *f + 1; /* the size must be at the start of the features string */
+ while ((p = strstr(p, o)) != NULL) {
+ if (isspace(*(p - 1)) &&
+ (isspace(*(p + d)) || *(p + d) == '\0'))
+ break;
+ p += d;
+ }
+ if (!p)
return 0;
/* Get feature count */
c = strtoul(*f, &e, 10);
- if (*f == e)
- /* parse error */
+ if (*f == e || !isspace(*e)) {
+ condlog(0, "parse error in feature string \"%s\"", *f);
return 1;
-
- /* Normalize features */
- while (*o == ' ') {
- o++;
}
- /* Just spaces, return */
- if (*o == '\0')
- return 0;
- q = o + strlen(o);
- while (*q == ' ')
- q--;
- d = (int)(q - o);
/* Update feature count */
c--;
q = o;
- while (q[0] != '\0') {
- if (q[0] == ' ' && q[1] != ' ' && q[1] != '\0')
+ while (*q != '\0') {
+ if (isspace(*q) && !isspace(*(q + 1)) && *(q + 1) != '\0')
c--;
q++;
}
@@ -684,15 +687,8 @@ int remove_feature(char **f, const char *o)
goto out;
}
- /* Search feature to be removed */
- e = strstr(*f, o);
- if (!e)
- /* Not found, return */
- return 0;
-
/* Update feature count space */
- l = strlen(*f) - d;
- n = MALLOC(l + 1);
+ n = MALLOC(strlen(*f) - d + 1);
if (!n)
return 1;
@@ -702,36 +698,16 @@ int remove_feature(char **f, const char *o)
* Copy existing features up to the feature
* about to be removed
*/
- p = strchr(*f, ' ');
- if (!p) {
- /* Internal error, feature string inconsistent */
- FREE(n);
- return 1;
- }
- while (*p == ' ')
- p++;
- p--;
- if (e != p) {
- do {
- e--;
- d++;
- } while (*e == ' ');
- e++; d--;
- strncat(n, p, (size_t)(e - p));
- p += (size_t)(e - p);
- }
+ strncat(n, e, (size_t)(p - e));
/* Skip feature to be removed */
p += d;
-
/* Copy remaining features */
- if (strlen(p)) {
- while (*p == ' ')
- p++;
- if (strlen(p)) {
- p--;
- strcat(n, p);
- }
- }
+ while (isspace(*p))
+ p++;
+ if (*p != '\0')
+ strcat(n, p);
+ else
+ strchop(n);
out:
FREE(*f);

View File

@ -0,0 +1,107 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:38 -0500
Subject: [PATCH] libmultipath: cleanup add_feature
add_feature() didn't correctly handle feature strings that used
whitespace other than spaces, which the kernel allows. It also didn't
allow adding features with multiple tokens. When it looked to see if the
feature string to be added already existed, it didn't check if the match
was part of a larger token. Finally, it did unnecessary work. By using
asprintf() to create the string, the function can be signifcantly
simplified.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/structs.c | 49 +++++++++++++++++++++---------------------
1 file changed, 24 insertions(+), 25 deletions(-)
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index 471087e2..84f9c959 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -572,23 +572,33 @@ int add_feature(char **f, const char *n)
{
int c = 0, d, l;
char *e, *t;
+ const char *p;
if (!f)
return 1;
/* Nothing to do */
- if (!n || *n == '0')
+ if (!n || *n == '\0')
return 0;
- if (strchr(n, ' ') != NULL) {
- condlog(0, "internal error: feature \"%s\" contains spaces", n);
+ l = strlen(n);
+ if (isspace(*n) || isspace(*(n + l - 1))) {
+ condlog(0, "internal error: feature \"%s\" has leading or trailing spaces", n);
return 1;
}
+ p = n;
+ d = 1;
+ while (*p != '\0') {
+ if (isspace(*p) && !isspace(*(p + 1)) && *(p + 1) != '\0')
+ d++;
+ p++;
+ }
+
/* default feature is null */
if(!*f)
{
- l = asprintf(&t, "1 %s", n);
+ l = asprintf(&t, "%0d %s", d, n);
if(l == -1)
return 1;
@@ -597,35 +607,24 @@ int add_feature(char **f, const char *n)
}
/* Check if feature is already present */
- if (strstr(*f, n))
- return 0;
+ e = *f;
+ while ((e = strstr(e, n)) != NULL) {
+ if (isspace(*(e - 1)) &&
+ (isspace(*(e + l)) || *(e + l) == '\0'))
+ return 0;
+ e += l;
+ }
/* Get feature count */
c = strtoul(*f, &e, 10);
- if (*f == e || (*e != ' ' && *e != '\0')) {
+ if (*f == e || (!isspace(*e) && *e != '\0')) {
condlog(0, "parse error in feature string \"%s\"", *f);
return 1;
}
-
- /* Add 1 digit and 1 space */
- l = strlen(e) + strlen(n) + 2;
-
- c++;
- /* Check if we need more digits for feature count */
- for (d = c; d >= 10; d /= 10)
- l++;
-
- t = MALLOC(l + 1);
- if (!t)
+ c += d;
+ if (asprintf(&t, "%0d%s %s", c, e, n) < 0)
return 1;
- /* e: old feature string with leading space, or "" */
- if (*e == ' ')
- while (*(e + 1) == ' ')
- e++;
-
- snprintf(t, l + 1, "%0d%s %s", c, e, n);
-
FREE(*f);
*f = t;

View File

@ -0,0 +1,350 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:39 -0500
Subject: [PATCH] multipath tests: tests for adding and removing features
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
tests/Makefile | 2 +-
tests/features.c | 318 +++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 319 insertions(+), 1 deletion(-)
create mode 100644 tests/features.c
diff --git a/tests/Makefile b/tests/Makefile
index 77ff3249..914413b8 100644
--- a/tests/Makefile
+++ b/tests/Makefile
@@ -13,7 +13,7 @@ CFLAGS += $(BIN_CFLAGS) -I$(multipathdir) -I$(mpathcmddir) \
LIBDEPS += -L$(multipathdir) -lmultipath -lcmocka
TESTS := uevent parser util dmevents hwtable blacklist unaligned vpd pgpolicy \
- alias directio
+ alias directio features
.SILENT: $(TESTS:%=%.o)
.PRECIOUS: $(TESTS:%=%-test)
diff --git a/tests/features.c b/tests/features.c
new file mode 100644
index 00000000..1e2e6bff
--- /dev/null
+++ b/tests/features.c
@@ -0,0 +1,318 @@
+#include <stddef.h>
+#include <stdarg.h>
+#include <setjmp.h>
+#include <cmocka.h>
+
+#include "structs.h"
+#include "globals.c"
+
+static void test_af_null_features_ptr(void **state)
+{
+ assert_int_equal(add_feature(NULL, "test"), 1);
+}
+
+static void af_helper(const char *features_start, const char *addition,
+ const char *features_end, int result)
+{
+ char *f = NULL, *orig = NULL;
+
+ if (features_start) {
+ f = strdup(features_start);
+ assert_non_null(f);
+ orig = f;
+ }
+ assert_int_equal(add_feature(&f, addition), result);
+ if (result != 0 || features_end == NULL)
+ assert_ptr_equal(orig, f);
+ else
+ assert_string_equal(f, features_end);
+ free(f);
+}
+
+static void test_af_null_addition1(void **state)
+{
+ af_helper("0", NULL, NULL, 0);
+}
+
+static void test_af_null_addition2(void **state)
+{
+ af_helper("1 queue_if_no_path", NULL, NULL, 0);
+}
+
+static void test_af_empty_addition(void **state)
+{
+ af_helper("2 pg_init_retries 5", "", NULL, 0);
+}
+
+static void test_af_invalid_addition1(void **state)
+{
+ af_helper("2 pg_init_retries 5", " ", NULL, 1);
+}
+
+static void test_af_invalid_addition2(void **state)
+{
+ af_helper("2 pg_init_retries 5", "\tbad", NULL, 1);
+}
+
+static void test_af_invalid_addition3(void **state)
+{
+ af_helper("2 pg_init_retries 5", "bad ", NULL, 1);
+}
+
+static void test_af_invalid_addition4(void **state)
+{
+ af_helper("2 pg_init_retries 5", " bad ", NULL, 1);
+}
+
+static void test_af_null_features1(void **state)
+{
+ af_helper(NULL, "test", "1 test", 0);
+}
+
+static void test_af_null_features2(void **state)
+{
+ af_helper(NULL, "test\t more", "2 test\t more", 0);
+}
+
+static void test_af_null_features3(void **state)
+{
+ af_helper(NULL, "test\neven\tmore", "3 test\neven\tmore", 0);
+}
+
+static void test_af_already_exists1(void **state)
+{
+ af_helper("4 this is a test", "test", NULL, 0);
+}
+
+static void test_af_already_exists2(void **state)
+{
+ af_helper("5 contest testy intestine test retest", "test", NULL, 0);
+}
+
+static void test_af_almost_exists(void **state)
+{
+ af_helper("3 contest testy intestine", "test",
+ "4 contest testy intestine test", 0);
+}
+
+static void test_af_bad_features1(void **state)
+{
+ af_helper("bad", "test", NULL, 1);
+}
+
+static void test_af_bad_features2(void **state)
+{
+ af_helper("1bad", "test", NULL, 1);
+}
+
+static void test_af_add1(void **state)
+{
+ af_helper("0", "test", "1 test", 0);
+}
+
+static void test_af_add2(void **state)
+{
+ af_helper("0", "this is a test", "4 this is a test", 0);
+}
+
+static void test_af_add3(void **state)
+{
+ af_helper("1 features", "more values", "3 features more values", 0);
+}
+
+static void test_af_add4(void **state)
+{
+ af_helper("2 one\ttwo", "three\t four", "4 one\ttwo three\t four", 0);
+}
+
+static int test_add_features(void)
+{
+ const struct CMUnitTest tests[] = {
+ cmocka_unit_test(test_af_null_features_ptr),
+ cmocka_unit_test(test_af_null_addition1),
+ cmocka_unit_test(test_af_null_addition2),
+ cmocka_unit_test(test_af_empty_addition),
+ cmocka_unit_test(test_af_invalid_addition1),
+ cmocka_unit_test(test_af_invalid_addition2),
+ cmocka_unit_test(test_af_invalid_addition3),
+ cmocka_unit_test(test_af_invalid_addition4),
+ cmocka_unit_test(test_af_null_features1),
+ cmocka_unit_test(test_af_null_features2),
+ cmocka_unit_test(test_af_null_features3),
+ cmocka_unit_test(test_af_already_exists1),
+ cmocka_unit_test(test_af_already_exists2),
+ cmocka_unit_test(test_af_almost_exists),
+ cmocka_unit_test(test_af_bad_features1),
+ cmocka_unit_test(test_af_bad_features2),
+ cmocka_unit_test(test_af_add1),
+ cmocka_unit_test(test_af_add2),
+ cmocka_unit_test(test_af_add3),
+ cmocka_unit_test(test_af_add4),
+ };
+ return cmocka_run_group_tests(tests, NULL, NULL);
+}
+
+static void test_rf_null_features_ptr(void **state)
+{
+ assert_int_equal(remove_feature(NULL, "test"), 1);
+}
+
+static void test_rf_null_features(void **state)
+{
+ char *f = NULL;
+ assert_int_equal(remove_feature(&f, "test"), 1);
+}
+
+static void rf_helper(const char *features_start, const char *removal,
+ const char *features_end, int result)
+{
+ char *f = strdup(features_start);
+ char *orig = f;
+
+ assert_non_null(f);
+ assert_int_equal(remove_feature(&f, removal), result);
+ if (result != 0 || features_end == NULL)
+ assert_ptr_equal(orig, f);
+ else
+ assert_string_equal(f, features_end);
+ free(f);
+}
+
+static void test_rf_null_removal(void **state)
+{
+ rf_helper("1 feature", NULL, NULL, 0);
+}
+
+static void test_rf_empty_removal(void **state)
+{
+ rf_helper("1 feature", "", NULL, 0);
+}
+
+static void test_rf_invalid_removal1(void **state)
+{
+ rf_helper("1 feature", " ", NULL, 1);
+}
+
+static void test_rf_invalid_removal2(void **state)
+{
+ rf_helper("1 feature", " bad", NULL, 1);
+}
+
+static void test_rf_invalid_removal3(void **state)
+{
+ rf_helper("1 feature", "bad\n", NULL, 1);
+}
+
+static void test_rf_invalid_removal4(void **state)
+{
+ rf_helper("1 feature", "\tbad \n", NULL, 1);
+}
+
+static void test_rf_bad_features1(void **state)
+{
+ rf_helper("invalid feature test string", "test", NULL, 1);
+}
+
+static void test_rf_bad_features2(void **state)
+{
+ rf_helper("2no space test", "test", NULL, 1);
+}
+
+static void test_rf_missing_removal1(void **state)
+{
+ rf_helper("0", "test", NULL, 0);
+}
+
+static void test_rf_missing_removal2(void **state)
+{
+ rf_helper("1 detest", "test", NULL, 0);
+}
+
+static void test_rf_missing_removal3(void **state)
+{
+ rf_helper("4 testing one two three", "test", NULL, 0);
+}
+
+static void test_rf_missing_removal4(void **state)
+{
+ rf_helper("1 contestant", "test", NULL, 0);
+}
+
+static void test_rf_missing_removal5(void **state)
+{
+ rf_helper("3 testament protest detestable", "test", NULL, 0);
+}
+
+static void test_rf_remove_all_features1(void **state)
+{
+ rf_helper("1 test", "test", "0", 0);
+}
+
+static void test_rf_remove_all_features2(void **state)
+{
+ rf_helper("2 another\t test", "another\t test", "0", 0);
+}
+
+static void test_rf_remove1(void **state)
+{
+ rf_helper("2 feature1 feature2", "feature2", "1 feature1", 0);
+}
+
+static void test_rf_remove2(void **state)
+{
+ rf_helper("2 feature1 feature2", "feature1", "1 feature2", 0);
+}
+
+static void test_rf_remove3(void **state)
+{
+ rf_helper("3 test1 test\ttest2", "test", "2 test1 test2", 0);
+}
+
+static void test_rf_remove4(void **state)
+{
+ rf_helper("4 this\t is a test", "is a", "2 this\t test", 0);
+}
+
+static void test_rf_remove5(void **state)
+{
+ rf_helper("3 one more test", "more test", "1 one", 0);
+}
+
+static int test_remove_features(void)
+{
+ const struct CMUnitTest tests[] = {
+ cmocka_unit_test(test_rf_null_features_ptr),
+ cmocka_unit_test(test_rf_null_features),
+ cmocka_unit_test(test_rf_null_removal),
+ cmocka_unit_test(test_rf_empty_removal),
+ cmocka_unit_test(test_rf_invalid_removal1),
+ cmocka_unit_test(test_rf_invalid_removal2),
+ cmocka_unit_test(test_rf_invalid_removal3),
+ cmocka_unit_test(test_rf_invalid_removal4),
+ cmocka_unit_test(test_rf_bad_features1),
+ cmocka_unit_test(test_rf_bad_features2),
+ cmocka_unit_test(test_rf_missing_removal1),
+ cmocka_unit_test(test_rf_missing_removal2),
+ cmocka_unit_test(test_rf_missing_removal3),
+ cmocka_unit_test(test_rf_missing_removal4),
+ cmocka_unit_test(test_rf_missing_removal5),
+ cmocka_unit_test(test_rf_remove_all_features1),
+ cmocka_unit_test(test_rf_remove_all_features2),
+ cmocka_unit_test(test_rf_remove1),
+ cmocka_unit_test(test_rf_remove2),
+ cmocka_unit_test(test_rf_remove3),
+ cmocka_unit_test(test_rf_remove4),
+ cmocka_unit_test(test_rf_remove5),
+ };
+ return cmocka_run_group_tests(tests, NULL, NULL);
+}
+
+int main(void)
+{
+ int ret = 0;
+
+ ret += test_add_features();
+ ret += test_remove_features();
+
+ return ret;
+}

View File

@ -0,0 +1,212 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:40 -0500
Subject: [PATCH] libmultipath: fix queue_mode feature handling
device-mapper is not able to change the queue_mode on a table reload.
Make sure that when multipath sets up the map, both on regular reloads
and reconfigures, it keeps the queue_mode the same.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/configure.c | 4 +++
libmultipath/dmparser.c | 2 ++
libmultipath/propsel.c | 55 ++++++++++++++++++++++++++++++++++++++
libmultipath/structs.h | 7 +++++
libmultipath/util.c | 10 +++++++
libmultipath/util.h | 1 +
multipath/multipath.conf.5 | 7 +++--
7 files changed, 84 insertions(+), 2 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 6cad0468..287289f7 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -1102,6 +1102,7 @@ int coalesce_paths (struct vectors * vecs, vector newmp, char * refwwid,
uint64_t *size_mismatch_seen;
bool map_processed = false;
bool no_daemon = false;
+ struct multipath * cmpp;
/* ignore refwwid if it's empty */
if (refwwid && !strlen(refwwid))
@@ -1197,6 +1198,9 @@ int coalesce_paths (struct vectors * vecs, vector newmp, char * refwwid,
}
verify_paths(mpp, vecs);
+ cmpp = find_mp_by_wwid(curmp, mpp->wwid);
+ if (cmpp)
+ mpp->queue_mode = cmpp->queue_mode;
params[0] = '\0';
if (setup_map(mpp, params, PARAMS_SIZE, vecs)) {
remove_map(mpp, vecs, 0);
diff --git a/libmultipath/dmparser.c b/libmultipath/dmparser.c
index b856a07f..b9c4dabc 100644
--- a/libmultipath/dmparser.c
+++ b/libmultipath/dmparser.c
@@ -164,6 +164,8 @@ int disassemble_map(vector pathvec, char *params, struct multipath *mpp,
FREE(word);
}
+ mpp->queue_mode = strstr(mpp->features, "queue_mode bio") ?
+ QUEUE_MODE_BIO : QUEUE_MODE_RQ;
/*
* hwhandler
diff --git a/libmultipath/propsel.c b/libmultipath/propsel.c
index be79902f..3f119dd9 100644
--- a/libmultipath/propsel.c
+++ b/libmultipath/propsel.c
@@ -26,6 +26,7 @@
#include "propsel.h"
#include <inttypes.h>
#include <libudev.h>
+#include <ctype.h>
pgpolicyfn *pgpolicies[] = {
NULL,
@@ -413,6 +414,59 @@ void reconcile_features_with_options(const char *id, char **features, int* no_pa
}
}
+static void reconcile_features_with_queue_mode(struct multipath *mp)
+{
+ char *space = NULL, *val = NULL, *mode_str = NULL, *feat;
+ int features_mode = QUEUE_MODE_UNDEF;
+
+ if (!mp->features)
+ return;
+
+ pthread_cleanup_push(cleanup_free_ptr, &space);
+ pthread_cleanup_push(cleanup_free_ptr, &val);
+ pthread_cleanup_push(cleanup_free_ptr, &mode_str);
+
+ if (!(feat = strstr(mp->features, "queue_mode")) ||
+ feat == mp->features || !isspace(*(feat - 1)) ||
+ sscanf(feat, "queue_mode%m[ \f\n\r\t\v]%ms", &space, &val) != 2)
+ goto sync_mode;
+ if (asprintf(&mode_str, "queue_mode%s%s", space, val) < 0) {
+ condlog(1, "failed to allocate space for queue_mode feature string");
+ mode_str = NULL; /* value undefined on failure */
+ goto exit;
+ }
+
+ if (!strcmp(val, "rq") || !strcmp(val, "mq"))
+ features_mode = QUEUE_MODE_RQ;
+ else if (!strcmp(val, "bio"))
+ features_mode = QUEUE_MODE_BIO;
+ if (features_mode == QUEUE_MODE_UNDEF) {
+ condlog(2, "%s: ignoring invalid feature '%s'",
+ mp->alias, mode_str);
+ goto sync_mode;
+ }
+
+ if (mp->queue_mode == QUEUE_MODE_UNDEF)
+ mp->queue_mode = features_mode;
+ if (mp->queue_mode == features_mode)
+ goto exit;
+
+ condlog(2,
+ "%s: ignoring feature '%s' because queue_mode is set to '%s'",
+ mp->alias, mode_str,
+ (mp->queue_mode == QUEUE_MODE_RQ)? "rq" : "bio");
+
+sync_mode:
+ if (mode_str)
+ remove_feature(&mp->features, mode_str);
+ if (mp->queue_mode == QUEUE_MODE_BIO)
+ add_feature(&mp->features, "queue_mode bio");
+exit:
+ pthread_cleanup_pop(1);
+ pthread_cleanup_pop(1);
+ pthread_cleanup_pop(1);
+}
+
int select_features(struct config *conf, struct multipath *mp)
{
const char *origin;
@@ -428,6 +482,7 @@ out:
reconcile_features_with_options(mp->alias, &mp->features,
&mp->no_path_retry,
&mp->retain_hwhandler);
+ reconcile_features_with_queue_mode(mp);
condlog(3, "%s: features = \"%s\" %s", mp->alias, mp->features, origin);
return 0;
}
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 3ed5cfc1..9a404da7 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -187,6 +187,12 @@ enum max_sectors_kb_states {
MAX_SECTORS_KB_MIN = 4, /* can't be smaller than page size */
};
+enum queue_mode_states {
+ QUEUE_MODE_UNDEF = 0,
+ QUEUE_MODE_BIO,
+ QUEUE_MODE_RQ,
+};
+
enum scsi_protocol {
SCSI_PROTOCOL_FCP = 0, /* Fibre Channel */
SCSI_PROTOCOL_SPI = 1, /* parallel SCSI */
@@ -397,6 +403,7 @@ struct multipath {
int needs_paths_uevent;
int ghost_delay;
int ghost_delay_tick;
+ int queue_mode;
uid_t uid;
gid_t gid;
mode_t mode;
diff --git a/libmultipath/util.c b/libmultipath/util.c
index dd30a46e..e04d20ab 100644
--- a/libmultipath/util.c
+++ b/libmultipath/util.c
@@ -465,6 +465,16 @@ void free_scandir_result(struct scandir_result *res)
FREE(res->di);
}
+void cleanup_free_ptr(void *arg)
+{
+ void **p = arg;
+
+ if (p && *p) {
+ free(*p);
+ *p = NULL;
+ }
+}
+
void close_fd(void *arg)
{
close((long)arg);
diff --git a/libmultipath/util.h b/libmultipath/util.h
index ce277680..f898c829 100644
--- a/libmultipath/util.h
+++ b/libmultipath/util.h
@@ -44,6 +44,7 @@ void set_max_fds(rlim_t max_fds);
pthread_cleanup_push(((void (*)(void *))&f), (arg))
void close_fd(void *arg);
+void cleanup_free_ptr(void *arg);
void cleanup_mutex(void *arg);
struct scandir_result {
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index 8e418372..61d2712b 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -458,8 +458,11 @@ precedence. See KNOWN ISSUES.
<mode> can be \fIbio\fR, \fIrq\fR or \fImq\fR, which corresponds to
bio-based, request-based, and block-multiqueue (blk-mq) request-based,
respectively.
-The default depends on the kernel parameter \fBdm_mod.use_blk_mq\fR. It is
-\fImq\fR if the latter is set, and \fIrq\fR otherwise.
+Before kernel 4.20 The default depends on the kernel parameter
+\fBdm_mod.use_blk_mq\fR. It is \fImq\fR if the latter is set, and \fIrq\fR
+otherwise. Since kernel 4.20, \fIrq\fR and \fImq\fR both correspond to
+block-multiqueue. Once a multipath device has been created, its queue_mode
+cannot be changed.
.TP
The default is: \fB<unset>\fR
.RE

View File

@ -0,0 +1,290 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:41 -0500
Subject: [PATCH] multipath tests: tests for reconcile_features_with_queue_mode
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
tests/Makefile | 2 +
tests/features.c | 232 ++++++++++++++++++++++++++++++++++++++++++++++-
2 files changed, 233 insertions(+), 1 deletion(-)
diff --git a/tests/Makefile b/tests/Makefile
index 914413b8..f3e49487 100644
--- a/tests/Makefile
+++ b/tests/Makefile
@@ -29,6 +29,7 @@ endif
ifneq ($(DIO_TEST_DEV),)
directio-test_FLAGS := -DDIO_TEST_DEV=\"$(DIO_TEST_DEV)\"
endif
+features-test_FLAGS := -I$(multipathdir)/nvme
# test-specific linker flags
# XYZ-test_TESTDEPS: test libraries containing __wrap_xyz functions
@@ -53,6 +54,7 @@ alias-test_LIBDEPS := -lpthread -ldl
ifneq ($(DIO_TEST_DEV),)
directio-test_LIBDEPS := -laio
endif
+features-test_LIBDEPS := -ludev -lpthread
%.o: %.c
$(CC) $(CFLAGS) $($*-test_FLAGS) -c -o $@ $<
diff --git a/tests/features.c b/tests/features.c
index 1e2e6bff..01fbccb7 100644
--- a/tests/features.c
+++ b/tests/features.c
@@ -1,9 +1,10 @@
+#define _GNU_SOURCE
#include <stddef.h>
#include <stdarg.h>
#include <setjmp.h>
#include <cmocka.h>
-#include "structs.h"
+#include "../libmultipath/propsel.c"
#include "globals.c"
static void test_af_null_features_ptr(void **state)
@@ -307,12 +308,241 @@ static int test_remove_features(void)
return cmocka_run_group_tests(tests, NULL, NULL);
}
+static void test_cf_null_features(void **state)
+{
+ struct multipath mp = {
+ .alias = "test",
+ };
+ reconcile_features_with_queue_mode(&mp);
+ assert_null(mp.features);
+}
+
+static void cf_helper(const char *features_start, const char *features_end,
+ int queue_mode_start, int queue_mode_end)
+{
+ struct multipath mp = {
+ .alias = "test",
+ .features = strdup(features_start),
+ .queue_mode = queue_mode_start,
+ };
+ char *orig = mp.features;
+
+ assert_non_null(orig);
+ reconcile_features_with_queue_mode(&mp);
+ if (!features_end)
+ assert_ptr_equal(orig, mp.features);
+ else
+ assert_string_equal(mp.features, features_end);
+ free(mp.features);
+ assert_int_equal(mp.queue_mode, queue_mode_end);
+}
+
+static void test_cf_unset_unset1(void **state)
+{
+ cf_helper("0", NULL, QUEUE_MODE_UNDEF, QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_unset_unset2(void **state)
+{
+ cf_helper("1 queue_mode", NULL, QUEUE_MODE_UNDEF, QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_unset_unset3(void **state)
+{
+ cf_helper("queue_mode", NULL, QUEUE_MODE_UNDEF, QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_unset_unset4(void **state)
+{
+ cf_helper("2 queue_model bio", NULL, QUEUE_MODE_UNDEF,
+ QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_unset_unset5(void **state)
+{
+ cf_helper("1 queue_if_no_path", NULL, QUEUE_MODE_UNDEF,
+ QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_invalid_unset1(void **state)
+{
+ cf_helper("2 queue_mode biop", "0", QUEUE_MODE_UNDEF, QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_invalid_unset2(void **state)
+{
+ cf_helper("3 queue_mode rqs queue_if_no_path", "1 queue_if_no_path",
+ QUEUE_MODE_UNDEF, QUEUE_MODE_UNDEF);
+}
+
+static void test_cf_rq_unset1(void **state)
+{
+ cf_helper("2 queue_mode rq", NULL, QUEUE_MODE_UNDEF, QUEUE_MODE_RQ);
+}
+
+static void test_cf_rq_unset2(void **state)
+{
+ cf_helper("2 queue_mode mq", NULL, QUEUE_MODE_UNDEF, QUEUE_MODE_RQ);
+}
+
+static void test_cf_bio_unset(void **state)
+{
+ cf_helper("2 queue_mode bio", NULL, QUEUE_MODE_UNDEF, QUEUE_MODE_BIO);
+}
+
+static void test_cf_unset_bio1(void **state)
+{
+ cf_helper("1 queue_if_no_path", "3 queue_if_no_path queue_mode bio",
+ QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_unset_bio2(void **state)
+{
+ cf_helper("0", "2 queue_mode bio", QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_unset_bio3(void **state)
+{
+ cf_helper("2 pg_init_retries 50", "4 pg_init_retries 50 queue_mode bio",
+ QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_invalid_bio1(void **state)
+{
+ cf_helper("2 queue_mode bad", "2 queue_mode bio",
+ QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_invalid_bio2(void **state)
+{
+ cf_helper("3 queue_if_no_path queue_mode\tbad", "3 queue_if_no_path queue_mode bio",
+ QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_bio_bio1(void **state)
+{
+ cf_helper("2 queue_mode bio", NULL, QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_bio_bio2(void **state)
+{
+ cf_helper("3 queue_if_no_path queue_mode bio", NULL, QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_bio_bio3(void **state)
+{
+ cf_helper("3 queue_mode\nbio queue_if_no_path", NULL, QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_bio_rq1(void **state)
+{
+ cf_helper("2\nqueue_mode\tbio", "0", QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_bio_rq2(void **state)
+{
+ cf_helper("3 queue_if_no_path\nqueue_mode bio", "1 queue_if_no_path",
+ QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_bio_rq3(void **state)
+{
+ cf_helper("4 queue_mode bio pg_init_retries 20", "2 pg_init_retries 20",
+ QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_unset_rq1(void **state)
+{
+ cf_helper("0", NULL, QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_unset_rq2(void **state)
+{
+ cf_helper("2 pg_init_retries 15", NULL, QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_invalid_rq1(void **state)
+{
+ cf_helper("2 queue_mode bionic", "0", QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_invalid_rq2(void **state)
+{
+ cf_helper("3 queue_mode b\nqueue_if_no_path", "1 queue_if_no_path",
+ QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_rq_rq1(void **state)
+{
+ cf_helper("2 queue_mode rq", NULL, QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_rq_rq2(void **state)
+{
+ cf_helper("3 queue_mode\t \trq\nqueue_if_no_path", NULL, QUEUE_MODE_RQ, QUEUE_MODE_RQ);
+}
+
+static void test_cf_rq_bio1(void **state)
+{
+ cf_helper("2 queue_mode rq", "2 queue_mode bio", QUEUE_MODE_BIO,
+ QUEUE_MODE_BIO);
+}
+
+static void test_cf_rq_bio2(void **state)
+{
+ cf_helper("3 queue_if_no_path\nqueue_mode rq", "3 queue_if_no_path queue_mode bio", QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static void test_cf_rq_bio3(void **state)
+{
+ cf_helper("3 queue_mode rq\nqueue_if_no_path", "3 queue_if_no_path queue_mode bio", QUEUE_MODE_BIO, QUEUE_MODE_BIO);
+}
+
+static int test_reconcile_features(void)
+{
+ const struct CMUnitTest tests[] = {
+ cmocka_unit_test(test_cf_null_features),
+ cmocka_unit_test(test_cf_unset_unset1),
+ cmocka_unit_test(test_cf_unset_unset2),
+ cmocka_unit_test(test_cf_unset_unset3),
+ cmocka_unit_test(test_cf_unset_unset4),
+ cmocka_unit_test(test_cf_unset_unset5),
+ cmocka_unit_test(test_cf_invalid_unset1),
+ cmocka_unit_test(test_cf_invalid_unset2),
+ cmocka_unit_test(test_cf_rq_unset1),
+ cmocka_unit_test(test_cf_rq_unset2),
+ cmocka_unit_test(test_cf_bio_unset),
+ cmocka_unit_test(test_cf_unset_bio1),
+ cmocka_unit_test(test_cf_unset_bio2),
+ cmocka_unit_test(test_cf_unset_bio3),
+ cmocka_unit_test(test_cf_invalid_bio1),
+ cmocka_unit_test(test_cf_invalid_bio2),
+ cmocka_unit_test(test_cf_bio_bio1),
+ cmocka_unit_test(test_cf_bio_bio2),
+ cmocka_unit_test(test_cf_bio_bio3),
+ cmocka_unit_test(test_cf_bio_rq1),
+ cmocka_unit_test(test_cf_bio_rq2),
+ cmocka_unit_test(test_cf_bio_rq3),
+ cmocka_unit_test(test_cf_unset_rq1),
+ cmocka_unit_test(test_cf_unset_rq2),
+ cmocka_unit_test(test_cf_invalid_rq1),
+ cmocka_unit_test(test_cf_invalid_rq2),
+ cmocka_unit_test(test_cf_rq_rq1),
+ cmocka_unit_test(test_cf_rq_rq2),
+ cmocka_unit_test(test_cf_rq_bio1),
+ cmocka_unit_test(test_cf_rq_bio2),
+ cmocka_unit_test(test_cf_rq_bio3),
+ };
+ return cmocka_run_group_tests(tests, NULL, NULL);
+}
+
int main(void)
{
int ret = 0;
ret += test_add_features();
ret += test_remove_features();
+ ret += test_reconcile_features();
return ret;
}

View File

@ -0,0 +1,143 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:42 -0500
Subject: [PATCH] libmultipath: prepare proto_id for use by non-scsi devivces
Make sure that when we are checking for a scsi protocol, we are first
checking that we are working with a scsi path.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/configure.c | 9 +++++----
libmultipath/discovery.c | 13 ++++++++-----
libmultipath/print.c | 6 ++++--
libmultipath/structs.c | 2 +-
libmultipath/structs.h | 4 +++-
multipathd/fpin_handlers.c | 2 +-
6 files changed, 22 insertions(+), 14 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 287289f7..8e1bc488 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -223,10 +223,11 @@ int rr_optimize_path_order(struct pathgroup *pgp)
total_paths = VECTOR_SIZE(pgp->paths);
vector_foreach_slot(pgp->paths, pp, i) {
- if (pp->sg_id.proto_id != SCSI_PROTOCOL_FCP &&
- pp->sg_id.proto_id != SCSI_PROTOCOL_SAS &&
- pp->sg_id.proto_id != SCSI_PROTOCOL_ISCSI &&
- pp->sg_id.proto_id != SCSI_PROTOCOL_SRP) {
+ if (pp->bus != SYSFS_BUS_SCSI ||
+ (pp->sg_id.proto_id != SCSI_PROTOCOL_FCP &&
+ pp->sg_id.proto_id != SCSI_PROTOCOL_SAS &&
+ pp->sg_id.proto_id != SCSI_PROTOCOL_ISCSI &&
+ pp->sg_id.proto_id != SCSI_PROTOCOL_SRP)) {
/* return success as default path order
* is maintained in path group
*/
diff --git a/libmultipath/discovery.c b/libmultipath/discovery.c
index 36cc389e..5f4e0794 100644
--- a/libmultipath/discovery.c
+++ b/libmultipath/discovery.c
@@ -468,10 +468,11 @@ int sysfs_get_host_adapter_name(const struct path *pp, char *adapter_name)
proto_id = pp->sg_id.proto_id;
- if (proto_id != SCSI_PROTOCOL_FCP &&
- proto_id != SCSI_PROTOCOL_SAS &&
- proto_id != SCSI_PROTOCOL_ISCSI &&
- proto_id != SCSI_PROTOCOL_SRP) {
+ if (pp->bus != SYSFS_BUS_SCSI ||
+ (proto_id != SCSI_PROTOCOL_FCP &&
+ proto_id != SCSI_PROTOCOL_SAS &&
+ proto_id != SCSI_PROTOCOL_ISCSI &&
+ proto_id != SCSI_PROTOCOL_SRP)) {
return 1;
}
/* iscsi doesn't have adapter info in sysfs
@@ -1722,8 +1723,10 @@ sysfs_pathinfo(struct path * pp, vector hwtable)
pp->bus = SYSFS_BUS_CCISS;
if (!strncmp(pp->dev,"dasd", 4))
pp->bus = SYSFS_BUS_CCW;
- if (!strncmp(pp->dev,"sd", 2))
+ if (!strncmp(pp->dev,"sd", 2)) {
pp->bus = SYSFS_BUS_SCSI;
+ pp->sg_id.proto_id = SCSI_PROTOCOL_UNSPEC;
+ }
if (!strncmp(pp->dev,"nvme", 4))
pp->bus = SYSFS_BUS_NVME;
diff --git a/libmultipath/print.c b/libmultipath/print.c
index 8a6fbe83..8a85df66 100644
--- a/libmultipath/print.c
+++ b/libmultipath/print.c
@@ -558,7 +558,8 @@ snprint_host_attr (char * buff, size_t len, const struct path * pp, char *attr)
const char *value = NULL;
int ret;
- if (pp->sg_id.proto_id != SCSI_PROTOCOL_FCP)
+ if (pp->bus != SYSFS_BUS_SCSI ||
+ pp->sg_id.proto_id != SCSI_PROTOCOL_FCP)
return snprintf(buff, len, "[undef]");
sprintf(host_id, "host%d", pp->sg_id.host_no);
host_dev = udev_device_new_from_subsystem_sysname(udev, "fc_host",
@@ -597,7 +598,8 @@ snprint_tgt_wwpn (char * buff, size_t len, const struct path * pp)
const char *value = NULL;
int ret;
- if (pp->sg_id.proto_id != SCSI_PROTOCOL_FCP)
+ if (pp->bus != SYSFS_BUS_SCSI ||
+ pp->sg_id.proto_id != SCSI_PROTOCOL_FCP)
return snprintf(buff, len, "[undef]");
sprintf(rport_id, "rport-%d:%d-%d",
pp->sg_id.host_no, pp->sg_id.channel, pp->sg_id.transport_id);
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index 84f9c959..1122cfae 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -115,7 +115,7 @@ alloc_path (void)
pp->sg_id.channel = -1;
pp->sg_id.scsi_id = -1;
pp->sg_id.lun = -1;
- pp->sg_id.proto_id = SCSI_PROTOCOL_UNSPEC;
+ pp->sg_id.proto_id = PROTOCOL_UNSET;
pp->fd = -1;
pp->tpgs = TPGS_UNDEF;
pp->priority = PRIO_UNDEF;
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 9a404da7..960ea024 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -193,6 +193,8 @@ enum queue_mode_states {
QUEUE_MODE_RQ,
};
+#define PROTOCOL_UNSET -1
+
enum scsi_protocol {
SCSI_PROTOCOL_FCP = 0, /* Fibre Channel */
SCSI_PROTOCOL_SPI = 1, /* parallel SCSI */
@@ -294,7 +296,7 @@ struct sg_id {
int lun;
short h_cmd_per_lun;
short d_queue_depth;
- enum scsi_protocol proto_id;
+ int proto_id;
int transport_id;
};
diff --git a/multipathd/fpin_handlers.c b/multipathd/fpin_handlers.c
index b14366d7..599f2893 100644
--- a/multipathd/fpin_handlers.c
+++ b/multipathd/fpin_handlers.c
@@ -220,7 +220,7 @@ static int fpin_chk_wwn_setpath_marginal(uint16_t host_num, struct vectors *ve
vector_foreach_slot(vecs->pathvec, pp, k) {
/* Checks the host number and also for the SCSI FCP */
- if (pp->sg_id.proto_id != SCSI_PROTOCOL_FCP || host_num != pp->sg_id.host_no)
+ if (pp->bus != SYSFS_BUS_SCSI || pp->sg_id.proto_id != SCSI_PROTOCOL_FCP || host_num != pp->sg_id.host_no)
continue;
sprintf(rport_id, "rport-%d:%d-%d",
pp->sg_id.host_no, pp->sg_id.channel, pp->sg_id.transport_id);

View File

@ -0,0 +1,194 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:43 -0500
Subject: [PATCH] libmultipath: get nvme path transport protocol
Read the transport protocol from /sys/block/nvmeXnY/device/transport.
Update protocol_name[] and bus_protocol_id() to store the nvme protocol
names after the scsi protocol names.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/discovery.c | 18 ++++++++++++++++--
libmultipath/structs.c | 22 +++++++++++++++++-----
libmultipath/structs.h | 33 +++++++++++++++++++++------------
multipath/multipath.conf.5 | 10 +++++++---
4 files changed, 61 insertions(+), 22 deletions(-)
diff --git a/libmultipath/discovery.c b/libmultipath/discovery.c
index 5f4e0794..eb7a634b 100644
--- a/libmultipath/discovery.c
+++ b/libmultipath/discovery.c
@@ -1455,6 +1455,7 @@ nvme_sysfs_pathinfo (struct path * pp, vector hwtable)
struct udev_device *parent;
const char *attr_path = NULL;
const char *attr;
+ int i;
attr_path = udev_device_get_sysname(pp->udev);
if (!attr_path)
@@ -1476,6 +1477,18 @@ nvme_sysfs_pathinfo (struct path * pp, vector hwtable)
attr = udev_device_get_sysattr_value(parent, "cntlid");
pp->sg_id.channel = attr ? atoi(attr) : 0;
+ attr = udev_device_get_sysattr_value(parent, "transport");
+ if (attr) {
+ for (i = 0; i < NVME_PROTOCOL_UNSPEC; i++){
+ if (protocol_name[SYSFS_BUS_NVME + i] &&
+ !strcmp(attr,
+ protocol_name[SYSFS_BUS_NVME + i] + 5)) {
+ pp->sg_id.proto_id = i;
+ break;
+ }
+ }
+ }
+
snprintf(pp->vendor_id, SCSI_VENDOR_SIZE, "NVME");
snprintf(pp->product_id, PATH_PRODUCT_SIZE, "%s",
udev_device_get_sysattr_value(parent, "model"));
@@ -1727,9 +1740,10 @@ sysfs_pathinfo(struct path * pp, vector hwtable)
pp->bus = SYSFS_BUS_SCSI;
pp->sg_id.proto_id = SCSI_PROTOCOL_UNSPEC;
}
- if (!strncmp(pp->dev,"nvme", 4))
+ if (!strncmp(pp->dev,"nvme", 4)) {
pp->bus = SYSFS_BUS_NVME;
-
+ pp->sg_id.proto_id = NVME_PROTOCOL_UNSPEC;
+ }
switch (pp->bus) {
case SYSFS_BUS_SCSI:
return scsi_sysfs_pathinfo(pp, hwtable);
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index 1122cfae..7bdf9152 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -25,7 +25,6 @@ const char * const protocol_name[LAST_BUS_PROTOCOL_ID + 1] = {
[SYSFS_BUS_UNDEF] = "undef",
[SYSFS_BUS_CCW] = "ccw",
[SYSFS_BUS_CCISS] = "cciss",
- [SYSFS_BUS_NVME] = "nvme",
[SYSFS_BUS_SCSI + SCSI_PROTOCOL_FCP] = "scsi:fcp",
[SYSFS_BUS_SCSI + SCSI_PROTOCOL_SPI] = "scsi:spi",
[SYSFS_BUS_SCSI + SCSI_PROTOCOL_SSA] = "scsi:ssa",
@@ -37,6 +36,13 @@ const char * const protocol_name[LAST_BUS_PROTOCOL_ID + 1] = {
[SYSFS_BUS_SCSI + SCSI_PROTOCOL_ATA] = "scsi:ata",
[SYSFS_BUS_SCSI + SCSI_PROTOCOL_USB] = "scsi:usb",
[SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC] = "scsi:unspec",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_PCIE] = "nvme:pcie",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_RDMA] = "nvme:rdma",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_FC] = "nvme:fc",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_TCP] = "nvme:tcp",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_LOOP] = "nvme:loop",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_APPLE_NVME] = "nvme:apple-nvme",
+ [SYSFS_BUS_NVME + NVME_PROTOCOL_UNSPEC] = "nvme:unspec",
};
struct adapter_group *
@@ -716,11 +722,17 @@ out:
}
unsigned int bus_protocol_id(const struct path *pp) {
- if (!pp || pp->bus < 0 || pp->bus > SYSFS_BUS_SCSI)
+ if (!pp || pp->bus < 0 || pp->bus > SYSFS_BUS_NVME)
return SYSFS_BUS_UNDEF;
- if (pp->bus != SYSFS_BUS_SCSI)
+ if (pp->bus != SYSFS_BUS_SCSI && pp->bus != SYSFS_BUS_NVME)
return pp->bus;
- if ((int)pp->sg_id.proto_id < 0 || pp->sg_id.proto_id > SCSI_PROTOCOL_UNSPEC)
+ if (pp->sg_id.proto_id < 0)
return SYSFS_BUS_UNDEF;
- return SYSFS_BUS_SCSI + pp->sg_id.proto_id;
+ if (pp->bus == SYSFS_BUS_SCSI &&
+ pp->sg_id.proto_id > SCSI_PROTOCOL_UNSPEC)
+ return SYSFS_BUS_UNDEF;
+ if (pp->bus == SYSFS_BUS_NVME &&
+ pp->sg_id.proto_id > NVME_PROTOCOL_UNSPEC)
+ return SYSFS_BUS_UNDEF;
+ return pp->bus + pp->sg_id.proto_id;
}
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 960ea024..9130efb5 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -57,15 +57,6 @@ enum failback_mode {
FAILBACK_FOLLOWOVER
};
-/* SYSFS_BUS_SCSI should be last, see bus_protocol_id() */
-enum sysfs_buses {
- SYSFS_BUS_UNDEF,
- SYSFS_BUS_CCW,
- SYSFS_BUS_CCISS,
- SYSFS_BUS_NVME,
- SYSFS_BUS_SCSI,
-};
-
enum pathstates {
PSTATE_UNDEF,
PSTATE_FAILED,
@@ -207,14 +198,32 @@ enum scsi_protocol {
SCSI_PROTOCOL_ATA = 8,
SCSI_PROTOCOL_USB = 9, /* USB Attached SCSI (UAS), and others */
SCSI_PROTOCOL_UNSPEC = 0xa, /* No specific protocol */
+ SCSI_PROTOCOL_END = 0xb, /* offset of the next sysfs_buses entry */
+};
+
+/* values from /sys/class/nvme/nvmeX */
+enum nvme_protocol {
+ NVME_PROTOCOL_PCIE = 0,
+ NVME_PROTOCOL_RDMA = 1,
+ NVME_PROTOCOL_FC = 2,
+ NVME_PROTOCOL_TCP = 3,
+ NVME_PROTOCOL_LOOP = 4,
+ NVME_PROTOCOL_APPLE_NVME = 5,
+ NVME_PROTOCOL_UNSPEC = 6, /* unknown protocol */
+};
+
+enum sysfs_buses {
+ SYSFS_BUS_UNDEF,
+ SYSFS_BUS_CCW,
+ SYSFS_BUS_CCISS,
+ SYSFS_BUS_SCSI,
+ SYSFS_BUS_NVME = SYSFS_BUS_SCSI + SCSI_PROTOCOL_END,
};
/*
* Linear ordering of bus/protocol
- * This assumes that SYSFS_BUS_SCSI is last in enum sysfs_buses
- * SCSI is the only bus type for which we distinguish protocols.
*/
-#define LAST_BUS_PROTOCOL_ID (SYSFS_BUS_SCSI + SCSI_PROTOCOL_UNSPEC)
+#define LAST_BUS_PROTOCOL_ID (SYSFS_BUS_NVME + NVME_PROTOCOL_UNSPEC)
unsigned int bus_protocol_id(const struct path *pp);
extern const char * const protocol_name[];
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index 61d2712b..1f5a40b6 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -1369,7 +1369,9 @@ Regular expression for the protocol of a device to be excluded/included.
The protocol strings that multipath recognizes are \fIscsi:fcp\fR,
\fIscsi:spi\fR, \fIscsi:ssa\fR, \fIscsi:sbp\fR, \fIscsi:srp\fR,
\fIscsi:iscsi\fR, \fIscsi:sas\fR, \fIscsi:adt\fR, \fIscsi:ata\fR,
-\fIscsi:unspec\fR, \fIccw\fR, \fIcciss\fR, \fInvme\fR, and \fIundef\fR.
+\fIscsi:unspec\fR, \fInvme:pcie\fR, \fInvme:rdma\fR, \fInvme:fc\fR,
+\fInvme:tcp\fR, \fInvme:loop\fR, \fInvme:apple-nvme\fR, \fInvme:unspec\fR,
+\fIccw\fR, \fIcciss\fR, and \fIundef\fR.
The protocol that a path is using can be viewed by running
\fBmultipathd show paths format "%d %P"\fR
.RE
@@ -1757,8 +1759,10 @@ The protocol subsection recognizes the following mandatory attribute:
The protocol string of the path device. The possible values are \fIscsi:fcp\fR,
\fIscsi:spi\fR, \fIscsi:ssa\fR, \fIscsi:sbp\fR, \fIscsi:srp\fR,
\fIscsi:iscsi\fR, \fIscsi:sas\fR, \fIscsi:adt\fR, \fIscsi:ata\fR,
-\fIscsi:unspec\fR, \fIccw\fR, \fIcciss\fR, \fInvme\fR, and \fIundef\fR. This is
-\fBnot\fR a regular expression. the path device protcol string must match
+\fIscsi:unspec\fR, \fInvme:pcie\fR, \fInvme:rdma\fR, \fInvme:fc\fR,
+\fInvme:tcp\fR, \fInvme:loop\fR, \fInvme:apple-nvme\fR, \fInvme:unspec\fR,
+\fIccw\fR, \fIcciss\fR, and \fIundef\fR. This is
+\fBnot\fR a regular expression. the path device protocol string must match
exactly. The protocol that a path is using can be viewed by running
\fBmultipathd show paths format "%d %P"\fR
.LP

View File

@ -0,0 +1,111 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Fri, 7 Oct 2022 12:35:44 -0500
Subject: [PATCH] libmultipath: enforce queue_mode bio for nmve:tcp paths
nvme:tcp devices set BLK_MQ_F_BLOCKING (they are the only block devices
which multipath supports that do so), meaning that block_mq expects that
they can block at certain points while servicing a request. However,
due to the way device-mapper sets up its queue, it is not able to set
BLK_MQ_F_BLOCKING when it includes paths that set this flag. Patches
were written to address this issue but they were rejected upstream
https://lore.kernel.org/linux-block/YcH%2FE4JNag0QYYAa@infradead.org/T/#t
The proposed solution was to have multipath use the bio queue_mode for
multipath devices that include nvme:tcp paths.
Multipath devices now automatically add the "queue_mode bio" feature if
they include nvme:tcp paths. If a multipath devices was created with
"queue_mode rq", it will disallow the addition of nvme:tcp paths.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/configure.c | 17 ++++++++++++++++-
libmultipath/structs_vec.c | 7 +++++++
multipath/multipath.conf.5 | 4 +++-
3 files changed, 26 insertions(+), 2 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 8e1bc488..c341793c 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -297,6 +297,7 @@ int setup_map(struct multipath *mpp, char *params, int params_size,
struct vectors *vecs)
{
struct pathgroup * pgp;
+ struct path *pp;
struct config *conf;
int i, n_paths, marginal_pathgroups;
@@ -308,6 +309,14 @@ int setup_map(struct multipath *mpp, char *params, int params_size,
return 1;
}
+ /* Force QUEUE_MODE_BIO for maps with nvme:tcp paths */
+ vector_foreach_slot(mpp->paths, pp, i) {
+ if (pp->bus == SYSFS_BUS_NVME &&
+ pp->sg_id.proto_id == NVME_PROTOCOL_TCP) {
+ mpp->queue_mode = QUEUE_MODE_BIO;
+ break;
+ }
+ }
/*
* free features, selector, and hwhandler properties if they are being reused
*/
@@ -1161,6 +1170,13 @@ int coalesce_paths (struct vectors * vecs, vector newmp, char * refwwid,
continue;
}
+ cmpp = find_mp_by_wwid(curmp, pp1->wwid);
+ if (cmpp && cmpp->queue_mode == QUEUE_MODE_RQ &&
+ pp1->bus == SYSFS_BUS_NVME && pp1->sg_id.proto_id ==
+ NVME_PROTOCOL_TCP) {
+ orphan_path(pp1, "nvme:tcp path not allowed with request queue_mode multipath device");
+ continue;
+ }
/*
* at this point, we know we really got a new mp
*/
@@ -1199,7 +1215,6 @@ int coalesce_paths (struct vectors * vecs, vector newmp, char * refwwid,
}
verify_paths(mpp, vecs);
- cmpp = find_mp_by_wwid(curmp, mpp->wwid);
if (cmpp)
mpp->queue_mode = cmpp->queue_mode;
params[0] = '\0';
diff --git a/libmultipath/structs_vec.c b/libmultipath/structs_vec.c
index 8137ea21..24ac022e 100644
--- a/libmultipath/structs_vec.c
+++ b/libmultipath/structs_vec.c
@@ -68,6 +68,13 @@ int adopt_paths(vector pathvec, struct multipath *mpp)
pp->dev, mpp->alias);
continue;
}
+ if (mpp->queue_mode == QUEUE_MODE_RQ &&
+ pp->bus == SYSFS_BUS_NVME &&
+ pp->sg_id.proto_id == NVME_PROTOCOL_TCP) {
+ condlog(2, "%s: mulitpath device %s created with request queue_mode. Unable to add nvme:tcp paths",
+ pp->dev, mpp->alias);
+ continue;
+ }
condlog(3, "%s: ownership set to %s",
pp->dev, mpp->alias);
pp->mpp = mpp;
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index 1f5a40b6..cb07a62c 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -462,7 +462,9 @@ Before kernel 4.20 The default depends on the kernel parameter
\fBdm_mod.use_blk_mq\fR. It is \fImq\fR if the latter is set, and \fIrq\fR
otherwise. Since kernel 4.20, \fIrq\fR and \fImq\fR both correspond to
block-multiqueue. Once a multipath device has been created, its queue_mode
-cannot be changed.
+cannot be changed. \fInvme:tcp\fR paths are only supported in multipath
+devices with queue_mode set to \fIbio\fR. multipath will automatically
+set this when creating a device with \fInvme:tcp\fR paths.
.TP
The default is: \fB<unset>\fR
.RE

View File

@ -0,0 +1,26 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 15 Nov 2022 09:01:36 -0600
Subject: [PATCH] multipath: add historical-service-time to the man page
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
multipath/multipath.conf.5 | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5
index cb07a62c..d8a98435 100644
--- a/multipath/multipath.conf.5
+++ b/multipath/multipath.conf.5
@@ -205,6 +205,11 @@ of outstanding I/O to the path.
(Since 2.6.31 kernel) Choose the path for the next bunch of I/O based on the amount
of outstanding I/O to the path and its relative throughput.
.TP
+.I "historical-service-time 0"
+(Since 4.18.0-305.3.el8 kernel) Choose the path for the next bunch of I/O based
+on the estimation of future service time based on the history of previous I/O
+submitted to each path.
+.TP
The default is: \fBservice-time 0\fR
.RE
.

View File

@ -0,0 +1,215 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Wed, 16 Sep 2020 22:22:36 +0200
Subject: [PATCH] libmultipath: copy mpp->hwe from pp->hwe
Since f0462f0 ("libmultipath: use vector for for pp->hwe and mp->hwe"),
we've been trying to fix issues caused by paths getting freed and mpp->hwe
dangling. This approach couldn't work because we need mpp->hwe to persist,
even if all paths are removed from the map. Before f0462f0, a simple
assignment worked, because the lifetime of the hwe wasn't bound to the
path. But now, we need to copy the vector. It turns out that we need to set
mpp->hwe only in two places, add_map_with_path() and setup_map(), and
that the code is simplified overall.
Even now, it can happen that a map is added with add_map_without_paths(),
and has no paths. In that case, calling do_set_from_hwe() with a NULL
pointer is not a bug, so remove the message.
Fixes: f0462f0 ("libmultipath: use vector for for pp->hwe and mp->hwe")
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/configure.c | 8 ++++++++
libmultipath/propsel.c | 2 +-
libmultipath/structs.c | 15 ++++++++++++++
libmultipath/structs.h | 1 +
libmultipath/structs_vec.c | 41 +++++++++++++++++++-------------------
multipathd/main.c | 10 ----------
6 files changed, 45 insertions(+), 32 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index c341793c..6e06fea2 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -324,6 +324,14 @@ int setup_map(struct multipath *mpp, char *params, int params_size,
if (mpp->disable_queueing && VECTOR_SIZE(mpp->paths) != 0)
mpp->disable_queueing = 0;
+ /*
+ * If this map was created with add_map_without_path(),
+ * mpp->hwe might not be set yet.
+ */
+ if (!mpp->hwe)
+ extract_hwe_from_path(mpp);
+
+
/*
* properties selectors
*
diff --git a/libmultipath/propsel.c b/libmultipath/propsel.c
index 3f119dd9..bc5c27ab 100644
--- a/libmultipath/propsel.c
+++ b/libmultipath/propsel.c
@@ -66,7 +66,7 @@ do { \
__do_set_from_vec(struct hwentry, var, (src)->hwe, dest)
#define do_set_from_hwe(var, src, dest, msg) \
- if (__do_set_from_hwe(var, src, dest)) { \
+ if (src->hwe && __do_set_from_hwe(var, src, dest)) { \
origin = msg; \
goto out; \
}
diff --git a/libmultipath/structs.c b/libmultipath/structs.c
index 7bdf9152..5bb6caf8 100644
--- a/libmultipath/structs.c
+++ b/libmultipath/structs.c
@@ -242,6 +242,17 @@ alloc_multipath (void)
return mpp;
}
+void *set_mpp_hwe(struct multipath *mpp, const struct path *pp)
+{
+ if (!mpp || !pp || !pp->hwe)
+ return NULL;
+ if (mpp->hwe)
+ return mpp->hwe;
+ mpp->hwe = vector_convert(NULL, pp->hwe,
+ struct hwentry, identity);
+ return mpp->hwe;
+}
+
void free_multipath_attributes(struct multipath *mpp)
{
if (!mpp)
@@ -283,6 +294,10 @@ free_multipath (struct multipath * mpp, enum free_path_mode free_paths)
free_pathvec(mpp->paths, free_paths);
free_pgvec(mpp->pg, free_paths);
+ if (mpp->hwe) {
+ vector_free(mpp->hwe);
+ mpp->hwe = NULL;
+ }
FREE_PTR(mpp->mpcontext);
FREE(mpp);
}
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 9130efb5..44980b4e 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -499,6 +499,7 @@ struct host_group {
struct path * alloc_path (void);
struct pathgroup * alloc_pathgroup (void);
struct multipath * alloc_multipath (void);
+void *set_mpp_hwe(struct multipath *mpp, const struct path *pp);
void free_path (struct path *);
void free_pathvec (vector vec, enum free_path_mode free_paths);
void free_pathgroup (struct pathgroup * pgp, enum free_path_mode free_paths);
diff --git a/libmultipath/structs_vec.c b/libmultipath/structs_vec.c
index 24ac022e..9613252f 100644
--- a/libmultipath/structs_vec.c
+++ b/libmultipath/structs_vec.c
@@ -180,24 +180,24 @@ extract_hwe_from_path(struct multipath * mpp)
if (mpp->hwe || !mpp->paths)
return;
- condlog(3, "%s: searching paths for valid hwe", mpp->alias);
+ condlog(4, "%s: searching paths for valid hwe", mpp->alias);
/* doing this in two passes seems like paranoia to me */
vector_foreach_slot(mpp->paths, pp, i) {
- if (pp->state != PATH_UP)
- continue;
- if (pp->hwe) {
- mpp->hwe = pp->hwe;
- return;
- }
+ if (pp->state == PATH_UP && pp->hwe)
+ goto done;
}
vector_foreach_slot(mpp->paths, pp, i) {
- if (pp->state == PATH_UP)
- continue;
- if (pp->hwe) {
- mpp->hwe = pp->hwe;
- return;
- }
+ if (pp->state != PATH_UP && pp->hwe)
+ goto done;
}
+done:
+ if (i < VECTOR_SIZE(mpp->paths))
+ (void)set_mpp_hwe(mpp, pp);
+
+ if (mpp->hwe)
+ condlog(3, "%s: got hwe from path %s", mpp->alias, pp->dev);
+ else
+ condlog(2, "%s: no hwe found", mpp->alias);
}
int
@@ -445,9 +445,15 @@ struct multipath *add_map_with_path(struct vectors *vecs, struct path *pp,
conf = get_multipath_config();
mpp->mpe = find_mpe(conf->mptable, pp->wwid);
- mpp->hwe = pp->hwe;
put_multipath_config(conf);
+ /*
+ * We need to call this before select_alias(),
+ * because that accesses hwe properties.
+ */
+ if (pp->hwe && !set_mpp_hwe(mpp, pp))
+ goto out;
+
strcpy(mpp->wwid, pp->wwid);
find_existing_alias(mpp, vecs);
if (select_alias(conf, mpp))
@@ -497,12 +503,6 @@ int verify_paths(struct multipath *mpp, struct vectors *vecs)
vector_del_slot(mpp->paths, i);
i--;
- /* Make sure mpp->hwe doesn't point to freed memory.
- * We call extract_hwe_from_path() below to restore
- * mpp->hwe
- */
- if (mpp->hwe == pp->hwe)
- mpp->hwe = NULL;
if ((j = find_slot(vecs->pathvec,
(void *)pp)) != -1)
vector_del_slot(vecs->pathvec, j);
@@ -512,7 +512,6 @@ int verify_paths(struct multipath *mpp, struct vectors *vecs)
mpp->alias, pp->dev, pp->dev_t);
}
}
- extract_hwe_from_path(mpp);
return count;
}
diff --git a/multipathd/main.c b/multipathd/main.c
index cd68a9d2..769dcaee 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -1198,13 +1198,6 @@ ev_remove_path (struct path *pp, struct vectors * vecs, int need_do_map)
goto fail;
}
- /*
- * Make sure mpp->hwe doesn't point to freed memory
- * We call extract_hwe_from_path() below to restore mpp->hwe
- */
- if (mpp->hwe == pp->hwe)
- mpp->hwe = NULL;
-
if ((i = find_slot(mpp->paths, (void *)pp)) != -1)
vector_del_slot(mpp->paths, i);
@@ -1216,9 +1209,6 @@ ev_remove_path (struct path *pp, struct vectors * vecs, int need_do_map)
flush_map_nopaths(mpp, vecs))
goto out;
- if (mpp->hwe == NULL)
- extract_hwe_from_path(mpp);
-
if (setup_map(mpp, params, PARAMS_SIZE, vecs)) {
condlog(0, "%s: failed to setup map for"
" removal of path %s", mpp->alias, pp->dev);

View File

@ -0,0 +1,35 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 14 Dec 2022 15:38:19 -0600
Subject: [PATCH] libmultipath: don't leak memory on invalid strings
If set_path() or set_str_noslash() are called with a bad value, they
ignore it and continue to use the old value. But they weren't freeing
the bad value, causing a memory leak.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/dict.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/libmultipath/dict.c b/libmultipath/dict.c
index d7cd94a5..a8c9e989 100644
--- a/libmultipath/dict.c
+++ b/libmultipath/dict.c
@@ -169,6 +169,7 @@ set_path(vector strvec, void *ptr, const char *file, int line_nr)
if ((*str_ptr)[0] != '/'){
condlog(1, "%s line %d, %s is not an absolute path. Ignoring",
file, line_nr, *str_ptr);
+ free(*str_ptr);
*str_ptr = old_str;
} else
free(old_str);
@@ -189,6 +190,7 @@ set_str_noslash(vector strvec, void *ptr, const char *file, int line_nr)
if (strchr(*str_ptr, '/')) {
condlog(1, "%s line %d, %s cannot contain a slash. Ignoring",
file, line_nr, *str_ptr);
+ free(*str_ptr);
*str_ptr = old_str;
} else
free(old_str);

View File

@ -0,0 +1,195 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 14 Dec 2022 15:38:20 -0600
Subject: [PATCH] libmutipath: validate the argument count of config strings
The features, path_selector, and hardware_handler config options pass
their strings directly into the kernel. If users omit the argument
counts from these strings, or use the wrong value, the kernel's table
parsing gets completely messed up, and the error messages it prints
don't reflect what actully went wrong. To avoid messing up the
kernel table parsing, verify that these strings correctly set the
argument count to the number of arguments they have.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmultipath/dict.c | 110 ++++++++++++++++++++++++++++++++++++++++----
1 file changed, 101 insertions(+), 9 deletions(-)
diff --git a/libmultipath/dict.c b/libmultipath/dict.c
index a8c9e989..952b33d0 100644
--- a/libmultipath/dict.c
+++ b/libmultipath/dict.c
@@ -155,6 +155,58 @@ set_dir(vector strvec, void *ptr, const char *file, int line_nr)
return 0;
}
+static int
+set_arg_str(vector strvec, void *ptr, int count_idx, const char *file,
+ int line_nr)
+{
+ char **str_ptr = (char **)ptr;
+ char *old_str = *str_ptr;
+ const char * const spaces = " \f\r\t\v";
+ char *p, *end;
+ int idx = -1;
+ long int count = -1;
+
+ *str_ptr = set_value(strvec);
+ if (!*str_ptr) {
+ free(old_str);
+ return 1;
+ }
+ p = *str_ptr;
+ while (*p != '\0') {
+ p += strspn(p, spaces);
+ if (*p == '\0')
+ break;
+ idx += 1;
+ if (idx == count_idx) {
+ errno = 0;
+ count = strtol(p, &end, 10);
+ if (errno == ERANGE || end == p ||
+ !(isspace(*end) || *end == '\0')) {
+ count = -1;
+ break;
+ }
+ }
+ p += strcspn(p, spaces);
+ }
+ if (count < 0) {
+ condlog(1, "%s line %d, missing argument count for %s",
+ file, line_nr, (char*)VECTOR_SLOT(strvec, 0));
+ goto fail;
+ }
+ if (count != idx - count_idx) {
+ condlog(1, "%s line %d, invalid argument count for %s:, got '%ld' expected '%d'",
+ file, line_nr, (char*)VECTOR_SLOT(strvec, 0), count,
+ idx - count_idx);
+ goto fail;
+ }
+ free(old_str);
+ return 0;
+fail:
+ free(*str_ptr);
+ *str_ptr = old_str;
+ return 0;
+}
+
static int
set_path(vector strvec, void *ptr, const char *file, int line_nr)
{
@@ -337,6 +389,14 @@ def_ ## option ## _handler (struct config *conf, vector strvec, \
return set_int(strvec, &conf->option, minval, maxval, file, line_nr); \
}
+#define declare_def_arg_str_handler(option, count_idx) \
+static int \
+def_ ## option ## _handler (struct config *conf, vector strvec, \
+ const char *file, int line_nr) \
+{ \
+ return set_arg_str(strvec, &conf->option, count_idx, file, line_nr); \
+}
+
#define declare_def_snprint(option, function) \
static int \
snprint_def_ ## option (struct config *conf, char * buff, int len, \
@@ -389,6 +449,17 @@ hw_ ## option ## _handler (struct config *conf, vector strvec, \
return set_int(strvec, &hwe->option, minval, maxval, file, line_nr); \
}
+#define declare_hw_arg_str_handler(option, count_idx) \
+static int \
+hw_ ## option ## _handler (struct config *conf, vector strvec, \
+ const char *file, int line_nr) \
+{ \
+ struct hwentry * hwe = VECTOR_LAST_SLOT(conf->hwtable); \
+ if (!hwe) \
+ return 1; \
+ return set_arg_str(strvec, &hwe->option, count_idx, file, line_nr); \
+}
+
#define declare_hw_snprint(option, function) \
static int \
@@ -420,6 +491,16 @@ ovr_ ## option ## _handler (struct config *conf, vector strvec, \
file, line_nr); \
}
+#define declare_ovr_arg_str_handler(option, count_idx) \
+static int \
+ovr_ ## option ## _handler (struct config *conf, vector strvec, \
+ const char *file, int line_nr) \
+{ \
+ if (!conf->overrides) \
+ return 1; \
+ return set_arg_str(strvec, &conf->overrides->option, count_idx, file, line_nr); \
+}
+
#define declare_ovr_snprint(option, function) \
static int \
snprint_ovr_ ## option (struct config *conf, char * buff, int len, \
@@ -450,6 +531,17 @@ mp_ ## option ## _handler (struct config *conf, vector strvec, \
return set_int(strvec, &mpe->option, minval, maxval, file, line_nr); \
}
+#define declare_mp_arg_str_handler(option, count_idx) \
+static int \
+mp_ ## option ## _handler (struct config *conf, vector strvec, \
+ const char *file, int line_nr) \
+{ \
+ struct mpentry * mpe = VECTOR_LAST_SLOT(conf->mptable); \
+ if (!mpe) \
+ return 1; \
+ return set_arg_str(strvec, &mpe->option, count_idx, file, line_nr); \
+}
+
#define declare_mp_snprint(option, function) \
static int \
snprint_mp_ ## option (struct config *conf, char * buff, int len, \
@@ -634,13 +726,13 @@ snprint_def_marginal_pathgroups(struct config *conf, char *buff, int len,
}
-declare_def_handler(selector, set_str)
+declare_def_arg_str_handler(selector, 1)
declare_def_snprint_defstr(selector, print_str, DEFAULT_SELECTOR)
-declare_hw_handler(selector, set_str)
+declare_hw_arg_str_handler(selector, 1)
declare_hw_snprint(selector, print_str)
-declare_ovr_handler(selector, set_str)
+declare_ovr_arg_str_handler(selector, 1)
declare_ovr_snprint(selector, print_str)
-declare_mp_handler(selector, set_str)
+declare_mp_arg_str_handler(selector, 1)
declare_mp_snprint(selector, print_str)
static int snprint_uid_attrs(struct config *conf, char *buff, int len,
@@ -717,13 +809,13 @@ declare_hw_snprint(prio_args, print_str)
declare_mp_handler(prio_args, set_str)
declare_mp_snprint(prio_args, print_str)
-declare_def_handler(features, set_str)
+declare_def_arg_str_handler(features, 0)
declare_def_snprint_defstr(features, print_str, DEFAULT_FEATURES)
-declare_ovr_handler(features, set_str)
+declare_ovr_arg_str_handler(features, 0)
declare_ovr_snprint(features, print_str)
-declare_hw_handler(features, set_str)
+declare_hw_arg_str_handler(features, 0)
declare_hw_snprint(features, print_str)
-declare_mp_handler(features, set_str)
+declare_mp_arg_str_handler(features, 0)
declare_mp_snprint(features, print_str)
declare_def_handler(checker_name, set_str)
@@ -1894,7 +1986,7 @@ declare_hw_snprint(revision, print_str)
declare_hw_handler(bl_product, set_regex)
declare_hw_snprint(bl_product, print_str)
-declare_hw_handler(hwhandler, set_str)
+declare_hw_arg_str_handler(hwhandler, 0)
declare_hw_snprint(hwhandler, print_str)
/*

View File

@ -0,0 +1,45 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 31 Jan 2023 09:58:55 -0600
Subject: [PATCH] libmultipath: select resize action even if reload is forced
The ACT_RESIZE action is the same as the ACT_RELOAD action, except that
it flushes outstanding IO because the device size is changing and
the new size might be too small for some of the outstanding IO. If we've
detected a size change, and a forced reload is requested, we still need
to flush the IO because the reload will change the device size.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/configure.c | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 6e06fea2..ecf24f95 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -734,17 +734,18 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
return;
}
- if (force_reload) {
+ if (cmpp->size != mpp->size) {
mpp->force_udev_reload = 1;
- mpp->action = ACT_RELOAD;
- condlog(3, "%s: set ACT_RELOAD (forced by user)",
+ mpp->action = ACT_RESIZE;
+ condlog(3, "%s: set ACT_RESIZE (size change)",
mpp->alias);
return;
}
- if (cmpp->size != mpp->size) {
+
+ if (force_reload) {
mpp->force_udev_reload = 1;
- mpp->action = ACT_RESIZE;
- condlog(3, "%s: set ACT_RESIZE (size change)",
+ mpp->action = ACT_RELOAD;
+ condlog(3, "%s: set ACT_RELOAD (forced by user)",
mpp->alias);
return;
}

View File

@ -0,0 +1,67 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 31 Jan 2023 10:35:10 -0600
Subject: [PATCH] libmultipath: cleanup ACT_CREATE code in select_action
Combine the two separate blocks that set ACT_CREATE into one.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/configure.c | 38 +++++++++++++++++---------------------
1 file changed, 17 insertions(+), 21 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index ecf24f95..303d2380 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -693,33 +693,29 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
cmpp = find_mp_by_wwid(curmp, mpp->wwid);
cmpp_by_name = find_mp_by_alias(curmp, mpp->alias);
- if (!cmpp_by_name) {
- if (cmpp) {
- condlog(2, "%s: rename %s to %s", mpp->wwid,
- cmpp->alias, mpp->alias);
- strlcpy(mpp->alias_old, cmpp->alias, WWID_SIZE);
- mpp->action = ACT_RENAME;
- if (force_reload) {
- mpp->force_udev_reload = 1;
- mpp->action = ACT_FORCERENAME;
- }
- return;
+ if (!cmpp) {
+ if (cmpp_by_name) {
+ condlog(1, "%s: can't use alias \"%s\" used by %s, falling back to WWID",
+ mpp->wwid, mpp->alias, cmpp_by_name->wwid);
+ /* We can do this because wwid wasn't found */
+ free(mpp->alias);
+ mpp->alias = strdup(mpp->wwid);
}
mpp->action = ACT_CREATE;
- condlog(3, "%s: set ACT_CREATE (map does not exist)",
- mpp->alias);
+ condlog(3, "%s: set ACT_CREATE (map does not exist%s)",
+ mpp->alias, cmpp_by_name ? ", name changed" : "");
return;
}
- if (!cmpp) {
- condlog(1, "%s: can't use alias \"%s\" used by %s, falling back to WWID",
- mpp->wwid, mpp->alias, cmpp_by_name->wwid);
- /* We can do this because wwid wasn't found */
- free(mpp->alias);
- mpp->alias = strdup(mpp->wwid);
- mpp->action = ACT_CREATE;
- condlog(3, "%s: set ACT_CREATE (map does not exist, name changed)",
+ if (!cmpp_by_name) {
+ condlog(2, "%s: rename %s to %s", mpp->wwid, cmpp->alias,
mpp->alias);
+ strlcpy(mpp->alias_old, cmpp->alias, WWID_SIZE);
+ mpp->action = ACT_RENAME;
+ if (force_reload) {
+ mpp->force_udev_reload = 1;
+ mpp->action = ACT_FORCERENAME;
+ }
return;
}

View File

@ -0,0 +1,270 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 31 Jan 2023 12:00:31 -0600
Subject: [PATCH] libmultipath: keep renames from stopping other multipath
actions
If select_action() is called and a multipath device needs to be renamed,
the code currently checks if force_reload is set, and if so, does the
reload after the rename. But if force_reload isn't set, only the rename
happens, regardless of what other actions are needed. This can happen if
multipathd starts up and a device needs both a reload and a rename.
Make multipath check for resize, reload, and switch pathgroup along with
rename, and do both if necessary.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/configure.c | 92 +++++++++++++++++++++-------------------
libmultipath/configure.h | 4 +-
2 files changed, 51 insertions(+), 45 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 303d2380..65a0b208 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -690,6 +690,7 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
struct multipath * cmpp_by_name;
char * mpp_feat, * cmpp_feat;
+ mpp->action = ACT_NOTHING;
cmpp = find_mp_by_wwid(curmp, mpp->wwid);
cmpp_by_name = find_mp_by_alias(curmp, mpp->alias);
@@ -712,14 +713,8 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
mpp->alias);
strlcpy(mpp->alias_old, cmpp->alias, WWID_SIZE);
mpp->action = ACT_RENAME;
- if (force_reload) {
- mpp->force_udev_reload = 1;
- mpp->action = ACT_FORCERENAME;
- }
- return;
- }
-
- if (cmpp != cmpp_by_name) {
+ /* don't return here. Check for other needed actions */
+ } else if (cmpp != cmpp_by_name) {
condlog(2, "%s: unable to rename %s to %s (%s is used by %s)",
mpp->wwid, cmpp->alias, mpp->alias,
mpp->alias, cmpp_by_name->wwid);
@@ -727,12 +722,13 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
FREE(mpp->alias);
mpp->alias = STRDUP(cmpp->alias);
mpp->action = ACT_IMPOSSIBLE;
- return;
+ /* don't return here. Check for other needed actions */
}
if (cmpp->size != mpp->size) {
mpp->force_udev_reload = 1;
- mpp->action = ACT_RESIZE;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RESIZE_RENAME :
+ ACT_RESIZE;
condlog(3, "%s: set ACT_RESIZE (size change)",
mpp->alias);
return;
@@ -740,7 +736,8 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
if (force_reload) {
mpp->force_udev_reload = 1;
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (forced by user)",
mpp->alias);
return;
@@ -749,7 +746,8 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
if (mpp->no_path_retry != NO_PATH_RETRY_UNDEF &&
!!strstr(mpp->features, "queue_if_no_path") !=
!!strstr(cmpp->features, "queue_if_no_path")) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (no_path_retry change)",
mpp->alias);
return;
@@ -759,7 +757,8 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
(strlen(cmpp->hwhandler) != strlen(mpp->hwhandler) ||
strncmp(cmpp->hwhandler, mpp->hwhandler,
strlen(mpp->hwhandler)))) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (hwhandler change)",
mpp->alias);
return;
@@ -769,7 +768,8 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
!!strstr(mpp->features, "retain_attached_hw_handler") !=
!!strstr(cmpp->features, "retain_attached_hw_handler") &&
get_linux_version_code() < KERNEL_VERSION(4, 3, 0)) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (retain_hwhandler change)",
mpp->alias);
return;
@@ -783,9 +783,13 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
remove_feature(&cmpp_feat, "queue_if_no_path");
remove_feature(&cmpp_feat, "retain_attached_hw_handler");
if (strncmp(mpp_feat, cmpp_feat, PARAMS_SIZE)) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ?
+ ACT_RELOAD_RENAME : ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (features change)",
mpp->alias);
+ FREE(cmpp_feat);
+ FREE(mpp_feat);
+ return;
}
}
FREE(cmpp_feat);
@@ -793,44 +797,49 @@ select_action (struct multipath * mpp, vector curmp, int force_reload)
if (!cmpp->selector || strncmp(cmpp->selector, mpp->selector,
strlen(mpp->selector))) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (selector change)",
mpp->alias);
return;
}
if (cmpp->minio != mpp->minio) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (minio change, %u->%u)",
mpp->alias, cmpp->minio, mpp->minio);
return;
}
if (!cmpp->pg || VECTOR_SIZE(cmpp->pg) != VECTOR_SIZE(mpp->pg)) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (path group number change)",
mpp->alias);
return;
}
if (pgcmp(mpp, cmpp)) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_RELOAD_RENAME :
+ ACT_RELOAD;
condlog(3, "%s: set ACT_RELOAD (path group topology change)",
mpp->alias);
return;
}
if (cmpp->nextpg != mpp->bestpg) {
- mpp->action = ACT_SWITCHPG;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_SWITCHPG_RENAME :
+ ACT_SWITCHPG;
condlog(3, "%s: set ACT_SWITCHPG (next path group change)",
mpp->alias);
return;
}
if (!is_mpp_known_to_udev(cmpp)) {
- mpp->action = ACT_RELOAD;
+ mpp->action = mpp->action == ACT_RENAME ? ACT_SWITCHPG_RENAME :
+ ACT_SWITCHPG;
condlog(3, "%s: set ACT_RELOAD (udev device not initialized)",
mpp->alias);
return;
}
- mpp->action = ACT_NOTHING;
- condlog(3, "%s: set ACT_NOTHING (map unchanged)",
- mpp->alias);
+ if (mpp->action == ACT_NOTHING)
+ condlog(3, "%s: set ACT_NOTHING (map unchanged)", mpp->alias);
return;
}
@@ -924,6 +933,17 @@ int domap(struct multipath *mpp, char *params, int is_daemon)
mpp->action = ACT_RELOAD;
}
+ if (mpp->action == ACT_RENAME || mpp->action == ACT_SWITCHPG_RENAME ||
+ mpp->action == ACT_RELOAD_RENAME ||
+ mpp->action == ACT_RESIZE_RENAME) {
+ conf = get_multipath_config();
+ pthread_cleanup_push(put_multipath_config, conf);
+ r = dm_rename(mpp->alias_old, mpp->alias,
+ conf->partition_delim, mpp->skip_kpartx);
+ pthread_cleanup_pop(1);
+ if (r == DOMAP_FAIL)
+ return r;
+ }
switch (mpp->action) {
case ACT_REJECT:
case ACT_NOTHING:
@@ -931,6 +951,7 @@ int domap(struct multipath *mpp, char *params, int is_daemon)
return DOMAP_EXIST;
case ACT_SWITCHPG:
+ case ACT_SWITCHPG_RENAME:
dm_switchgroup(mpp->alias, mpp->bestpg);
/*
* we may have avoided reinstating paths because there where in
@@ -957,6 +978,7 @@ int domap(struct multipath *mpp, char *params, int is_daemon)
break;
case ACT_RELOAD:
+ case ACT_RELOAD_RENAME:
sysfs_set_max_sectors_kb(mpp, 1);
if (mpp->ghost_delay_tick > 0 && pathcount(mpp, PATH_UP))
mpp->ghost_delay_tick = 0;
@@ -964,6 +986,7 @@ int domap(struct multipath *mpp, char *params, int is_daemon)
break;
case ACT_RESIZE:
+ case ACT_RESIZE_RENAME:
sysfs_set_max_sectors_kb(mpp, 1);
if (mpp->ghost_delay_tick > 0 && pathcount(mpp, PATH_UP))
mpp->ghost_delay_tick = 0;
@@ -971,29 +994,10 @@ int domap(struct multipath *mpp, char *params, int is_daemon)
break;
case ACT_RENAME:
- conf = get_multipath_config();
- pthread_cleanup_push(put_multipath_config, conf);
- r = dm_rename(mpp->alias_old, mpp->alias,
- conf->partition_delim, mpp->skip_kpartx);
- pthread_cleanup_pop(1);
- break;
-
- case ACT_FORCERENAME:
- conf = get_multipath_config();
- pthread_cleanup_push(put_multipath_config, conf);
- r = dm_rename(mpp->alias_old, mpp->alias,
- conf->partition_delim, mpp->skip_kpartx);
- pthread_cleanup_pop(1);
- if (r) {
- sysfs_set_max_sectors_kb(mpp, 1);
- if (mpp->ghost_delay_tick > 0 &&
- pathcount(mpp, PATH_UP))
- mpp->ghost_delay_tick = 0;
- r = dm_addmap_reload(mpp, params, 0);
- }
break;
default:
+ r = DOMAP_FAIL;
break;
}
diff --git a/libmultipath/configure.h b/libmultipath/configure.h
index 5cf08d45..1a93f49d 100644
--- a/libmultipath/configure.h
+++ b/libmultipath/configure.h
@@ -18,9 +18,11 @@ enum actions {
ACT_RENAME,
ACT_CREATE,
ACT_RESIZE,
- ACT_FORCERENAME,
+ ACT_RELOAD_RENAME,
ACT_DRY_RUN,
ACT_IMPOSSIBLE,
+ ACT_RESIZE_RENAME,
+ ACT_SWITCHPG_RENAME,
};
/*

View File

@ -0,0 +1,136 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 20 Dec 2022 17:41:10 -0600
Subject: [PATCH] multipathd: make pr registration consistent
multipathd was inconsistent on what it did with persistent reservations
when a multipath device was created. If a multipath device with a
configured reservation key was created during configure(), multipathd
would try to read the registered keys using an active path. If it saw a
matching key, it would set the prflag, but not attempt to register the
key on any of the other paths. This means that if a new path had
appeared while multipathd was not running, it wouldn't register the key
on this path.
If the multipath device was created during ev_add_path(), multipathd
would used the added path to check if there was a matching key and if
there was, register the key only on the added path and then set the
prflag. This could be problematic if the device was created with
multiple paths, for instance because find_mutipaths was set to "yes" and
a second path just appeared. In this case, if the device happened to be
only registered on the second path, it would not get registered on the
first path.
If the multipath device was added to multipathd during a call to
ev_add_map(), multipathd wouldn't set the prflag or register the key on
any paths.
After a device was created with the prflag set, if a new path appeared
before the creation uevent, and multipathd was forced to delay adding
it, when it finally updated the multipath device, the key would be
registered on all paths, fixing any paths missed during creation.
However, if a new path appeared after the creation uevent, the key would
only be registered on that new path. Any paths that were missed on
creation would stay missed.
persistent key registration needs to be handled consistently. This
patch does so by making sure that however a multipath device is added to
multipathd, it will check to see if the configured key is registered. If
it is, multipathd will set the prflag and register the key on all the
currently active paths.
When a new path is added, multipathd will use it to check for active
keys, as before. But if it finds a matching key and prflag isn't
currently set, it will register the key on all paths.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipathd/main.c | 32 +++++++++++++++++++++++++++++---
1 file changed, 29 insertions(+), 3 deletions(-)
diff --git a/multipathd/main.c b/multipathd/main.c
index 769dcaee..d84027dc 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -511,6 +511,21 @@ flush_map_nopaths(struct multipath *mpp, struct vectors *vecs) {
return false;
}
+static void
+pr_register_active_paths(struct multipath *mpp)
+{
+ unsigned int i, j;
+ struct path *pp;
+ struct pathgroup *pgp;
+
+ vector_foreach_slot (mpp->pg, pgp, i) {
+ vector_foreach_slot (pgp->paths, pp, j) {
+ if ((pp->state == PATH_UP) || (pp->state == PATH_GHOST))
+ mpath_pr_event_handle(pp);
+ }
+ }
+}
+
static int
update_map (struct multipath *mpp, struct vectors *vecs, int new_map)
{
@@ -556,6 +571,11 @@ fail:
sync_map_state(mpp);
+ if (!mpp->prflag)
+ update_map_pr(mpp);
+ if (mpp->prflag)
+ pr_register_active_paths(mpp);
+
if (retries < 0)
condlog(0, "%s: failed reload in new map update", mpp->alias);
return 0;
@@ -1014,6 +1034,7 @@ ev_add_path (struct path * pp, struct vectors * vecs, int need_do_map)
int start_waiter = 0;
int ret;
int ro;
+ unsigned char prflag = 0;
/*
* need path UID to go any further
@@ -1056,6 +1077,8 @@ rescan:
verify_paths(mpp, vecs);
mpp->action = ACT_RELOAD;
+ prflag = mpp->prflag;
+ mpath_pr_event_handle(pp);
} else {
if (!should_multipath(pp, vecs->pathvec, vecs->mpvec)) {
orphan_path(pp, "only one path");
@@ -1074,9 +1097,6 @@ rescan:
goto fail; /* leave path added to pathvec */
}
- /* persistent reservation check*/
- mpath_pr_event_handle(pp);
-
/* ro check - if new path is ro, force map to be ro as well */
ro = sysfs_get_ro(pp);
if (ro == 1)
@@ -1140,6 +1160,10 @@ rescan:
sync_map_state(mpp);
if (retries >= 0) {
+ if (start_waiter)
+ update_map_pr(mpp);
+ if (mpp->prflag && !prflag)
+ pr_register_active_paths(mpp);
condlog(2, "%s [%s]: path added to devmap %s",
pp->dev, pp->dev_t, mpp->alias);
return 0;
@@ -2608,6 +2632,8 @@ configure (struct vectors * vecs)
if (remember_wwid(mpp->wwid) == 1)
trigger_paths_udev_change(mpp, true);
update_map_pr(mpp);
+ if (mpp->prflag)
+ pr_register_active_paths(mpp);
}
/*

View File

@ -0,0 +1,165 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 20 Dec 2022 17:41:11 -0600
Subject: [PATCH] libmultipath: make prflag an enum
In preparation for a future patch, make prflag an enum, and change the
reply of cli_getprstatus() to a string.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmpathpersist/mpath_persist.c | 2 +-
libmultipath/structs.h | 8 +++++++-
multipathd/cli_handlers.c | 16 +++++++++-------
multipathd/main.c | 14 +++++++-------
4 files changed, 24 insertions(+), 16 deletions(-)
diff --git a/libmpathpersist/mpath_persist.c b/libmpathpersist/mpath_persist.c
index d0744773..e361435a 100644
--- a/libmpathpersist/mpath_persist.c
+++ b/libmpathpersist/mpath_persist.c
@@ -949,7 +949,7 @@ int update_map_pr(struct multipath *mpp)
if (isFound)
{
- mpp->prflag = 1;
+ mpp->prflag = PRFLAG_SET;
condlog(2, "%s: prflag flag set.", mpp->alias );
}
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 44980b4e..8acea3cb 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -377,6 +377,12 @@ struct path {
typedef int (pgpolicyfn) (struct multipath *, vector);
+
+enum prflag_value {
+ PRFLAG_UNSET,
+ PRFLAG_SET,
+};
+
struct multipath {
char wwid[WWID_SIZE];
char alias_old[WWID_SIZE];
@@ -450,7 +456,7 @@ struct multipath {
int prkey_source;
struct be64 reservation_key;
uint8_t sa_flags;
- unsigned char prflag;
+ int prflag;
int all_tg_pt;
struct gen_multipath generic_mp;
bool fpin_must_reload;
diff --git a/multipathd/cli_handlers.c b/multipathd/cli_handlers.c
index 8b4bd187..6dbbb7e2 100644
--- a/multipathd/cli_handlers.c
+++ b/multipathd/cli_handlers.c
@@ -1407,6 +1407,10 @@ cli_shutdown (void * v, char ** reply, int * len, void * data)
int
cli_getprstatus (void * v, char ** reply, int * len, void * data)
{
+ static const char * const prflag_str[] = {
+ [PRFLAG_UNSET] = "unset\n",
+ [PRFLAG_SET] = "set\n",
+ };
struct multipath * mpp;
struct vectors * vecs = (struct vectors *)data;
char * param = get_keyparam(v, MAP);
@@ -1418,9 +1422,7 @@ cli_getprstatus (void * v, char ** reply, int * len, void * data)
if (!mpp)
return 1;
- condlog(3, "%s: prflag = %u", param, (unsigned int)mpp->prflag);
-
- *len = asprintf(reply, "%d", mpp->prflag);
+ *len = asprintf(reply, "%s", prflag_str[mpp->prflag]);
if (*len < 0)
return 1;
@@ -1443,8 +1445,8 @@ cli_setprstatus(void * v, char ** reply, int * len, void * data)
if (!mpp)
return 1;
- if (!mpp->prflag) {
- mpp->prflag = 1;
+ if (mpp->prflag != PRFLAG_SET) {
+ mpp->prflag = PRFLAG_SET;
condlog(2, "%s: prflag set", param);
}
@@ -1466,8 +1468,8 @@ cli_unsetprstatus(void * v, char ** reply, int * len, void * data)
if (!mpp)
return 1;
- if (mpp->prflag) {
- mpp->prflag = 0;
+ if (mpp->prflag != PRFLAG_UNSET) {
+ mpp->prflag = PRFLAG_UNSET;
condlog(2, "%s: prflag unset", param);
}
diff --git a/multipathd/main.c b/multipathd/main.c
index d84027dc..81bb0deb 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -571,9 +571,9 @@ fail:
sync_map_state(mpp);
- if (!mpp->prflag)
+ if (mpp->prflag == PRFLAG_UNSET)
update_map_pr(mpp);
- if (mpp->prflag)
+ if (mpp->prflag == PRFLAG_SET)
pr_register_active_paths(mpp);
if (retries < 0)
@@ -1034,7 +1034,7 @@ ev_add_path (struct path * pp, struct vectors * vecs, int need_do_map)
int start_waiter = 0;
int ret;
int ro;
- unsigned char prflag = 0;
+ unsigned char prflag = PRFLAG_UNSET;
/*
* need path UID to go any further
@@ -1162,7 +1162,7 @@ rescan:
if (retries >= 0) {
if (start_waiter)
update_map_pr(mpp);
- if (mpp->prflag && !prflag)
+ if (mpp->prflag == PRFLAG_SET && prflag == PRFLAG_UNSET)
pr_register_active_paths(mpp);
condlog(2, "%s [%s]: path added to devmap %s",
pp->dev, pp->dev_t, mpp->alias);
@@ -2306,7 +2306,7 @@ check_path (struct vectors * vecs, struct path * pp, unsigned int ticks)
}
if (newstate == PATH_UP || newstate == PATH_GHOST) {
- if (pp->mpp->prflag) {
+ if (pp->mpp->prflag == PRFLAG_SET) {
/*
* Check Persistent Reservation.
*/
@@ -2632,7 +2632,7 @@ configure (struct vectors * vecs)
if (remember_wwid(mpp->wwid) == 1)
trigger_paths_udev_change(mpp, true);
update_map_pr(mpp);
- if (mpp->prflag)
+ if (mpp->prflag == PRFLAG_SET)
pr_register_active_paths(mpp);
}
@@ -3438,7 +3438,7 @@ void * mpath_pr_event_handler_fn (void * pathp )
{
condlog(0,"%s: Reservation registration failed. Error: %d", pp->dev, ret);
}
- mpp->prflag = 1;
+ mpp->prflag = PRFLAG_SET;
free(param);
out:

View File

@ -0,0 +1,152 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Tue, 20 Dec 2022 17:41:12 -0600
Subject: [PATCH] multipathd: handle no active paths in update_map_pr
When a multipath device is first created, if it has a reservation key
configured, update_map_pr() will check for a matching key on the active
paths. If there were no active paths to check with, multipathd was
leaving mpp->prflag in PRFLAG_UNSET, as if there were no matching keys.
It's possible that when update_map_pr() is called, all the paths will be
in the PATH_PENDING state because the checkers haven't completed yet. In
this case, multipathd was treating the device as having no registered
keys without ever checking.
To solve this, multipath devices now start with prflag = PRFLAG_UNKNOWN.
It will remain in this state until multipathd actually tries to get the
registered keys down a path. If the map is in this state, it will check
newly active paths, and if it finds a matching key, it will register
the key down all active paths.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
libmpathpersist/mpath_persist.c | 8 ++++++++
libmultipath/structs.h | 1 +
multipathd/cli_handlers.c | 1 +
multipathd/main.c | 19 ++++++++++++++-----
4 files changed, 24 insertions(+), 5 deletions(-)
diff --git a/libmpathpersist/mpath_persist.c b/libmpathpersist/mpath_persist.c
index e361435a..440a329f 100644
--- a/libmpathpersist/mpath_persist.c
+++ b/libmpathpersist/mpath_persist.c
@@ -904,6 +904,7 @@ int update_map_pr(struct multipath *mpp)
if (!get_be64(mpp->reservation_key))
{
/* Nothing to do. Assuming pr mgmt feature is disabled*/
+ mpp->prflag = PRFLAG_UNSET;
condlog(4, "%s: reservation_key not set in multipath.conf",
mpp->alias);
return MPATH_PR_SUCCESS;
@@ -915,6 +916,13 @@ int update_map_pr(struct multipath *mpp)
condlog(0,"%s : failed to alloc resp in update_map_pr", mpp->alias);
return MPATH_PR_OTHER;
}
+ if (count_active_paths(mpp) == 0)
+ {
+ condlog(0,"%s: No available paths to check pr status",
+ mpp->alias);
+ return MPATH_PR_OTHER;
+ }
+ mpp->prflag = PRFLAG_UNSET;
ret = mpath_prin_activepath(mpp, MPATH_PRIN_RKEY_SA, resp, noisy);
if (ret != MPATH_PR_SUCCESS )
diff --git a/libmultipath/structs.h b/libmultipath/structs.h
index 8acea3cb..7f2fb944 100644
--- a/libmultipath/structs.h
+++ b/libmultipath/structs.h
@@ -379,6 +379,7 @@ typedef int (pgpolicyfn) (struct multipath *, vector);
enum prflag_value {
+ PRFLAG_UNKNOWN,
PRFLAG_UNSET,
PRFLAG_SET,
};
diff --git a/multipathd/cli_handlers.c b/multipathd/cli_handlers.c
index 6dbbb7e2..260b7a17 100644
--- a/multipathd/cli_handlers.c
+++ b/multipathd/cli_handlers.c
@@ -1408,6 +1408,7 @@ int
cli_getprstatus (void * v, char ** reply, int * len, void * data)
{
static const char * const prflag_str[] = {
+ [PRFLAG_UNKNOWN] = "unknown\n",
[PRFLAG_UNSET] = "unset\n",
[PRFLAG_SET] = "set\n",
};
diff --git a/multipathd/main.c b/multipathd/main.c
index 81bb0deb..e7dad6b9 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -571,7 +571,7 @@ fail:
sync_map_state(mpp);
- if (mpp->prflag == PRFLAG_UNSET)
+ if (mpp->prflag != PRFLAG_SET)
update_map_pr(mpp);
if (mpp->prflag == PRFLAG_SET)
pr_register_active_paths(mpp);
@@ -1162,7 +1162,7 @@ rescan:
if (retries >= 0) {
if (start_waiter)
update_map_pr(mpp);
- if (mpp->prflag == PRFLAG_SET && prflag == PRFLAG_UNSET)
+ if (mpp->prflag == PRFLAG_SET && prflag != PRFLAG_SET)
pr_register_active_paths(mpp);
condlog(2, "%s [%s]: path added to devmap %s",
pp->dev, pp->dev_t, mpp->alias);
@@ -2306,13 +2306,17 @@ check_path (struct vectors * vecs, struct path * pp, unsigned int ticks)
}
if (newstate == PATH_UP || newstate == PATH_GHOST) {
- if (pp->mpp->prflag == PRFLAG_SET) {
+ if (pp->mpp->prflag != PRFLAG_UNSET) {
+ int prflag = pp->mpp->prflag;
/*
* Check Persistent Reservation.
*/
condlog(2, "%s: checking persistent "
"reservation registration", pp->dev);
mpath_pr_event_handle(pp);
+ if (pp->mpp->prflag == PRFLAG_SET &&
+ prflag != PRFLAG_SET)
+ pr_register_active_paths(pp->mpp);
}
}
@@ -3386,6 +3390,7 @@ void * mpath_pr_event_handler_fn (void * pathp )
goto out;
}
+ mpp->prflag = PRFLAG_UNSET;
ret = prin_do_scsi_ioctl(pp->dev, MPATH_PRIN_RKEY_SA, resp, 0);
if (ret != MPATH_PR_SUCCESS )
{
@@ -3456,12 +3461,12 @@ int mpath_pr_event_handle(struct path *pp)
struct multipath * mpp;
if (pp->bus != SYSFS_BUS_SCSI)
- return 0;
+ goto no_pr;
mpp = pp->mpp;
if (!get_be64(mpp->reservation_key))
- return -1;
+ goto no_pr;
pthread_attr_init(&attr);
pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_JOINABLE);
@@ -3474,4 +3479,8 @@ int mpath_pr_event_handle(struct path *pp)
pthread_attr_destroy(&attr);
rc = pthread_join(thread, NULL);
return 0;
+
+no_pr:
+ pp->mpp->prflag = PRFLAG_UNSET;
+ return 0;
}

View File

@ -0,0 +1,68 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Thu, 2 Feb 2023 09:20:35 +0100
Subject: [PATCH] libmpathpersist: fix resource leak in update_map_pr()
The "no available paths" case would leak the memory resp points to.
Found by coverity.
Fixes: 50e2c16 ("multipathd: handle no active paths in update_map_pr")
Signed-off-by: Martin Wilck <mwilck@suse.com>
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmpathpersist/mpath_persist.c | 15 ++++++++-------
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/libmpathpersist/mpath_persist.c b/libmpathpersist/mpath_persist.c
index 440a329f..a4c1461f 100644
--- a/libmpathpersist/mpath_persist.c
+++ b/libmpathpersist/mpath_persist.c
@@ -899,7 +899,7 @@ int update_map_pr(struct multipath *mpp)
int noisy=0;
struct prin_resp *resp;
unsigned int i;
- int ret, isFound;
+ int ret = MPATH_PR_OTHER, isFound;
if (!get_be64(mpp->reservation_key))
{
@@ -920,7 +920,7 @@ int update_map_pr(struct multipath *mpp)
{
condlog(0,"%s: No available paths to check pr status",
mpp->alias);
- return MPATH_PR_OTHER;
+ goto out;
}
mpp->prflag = PRFLAG_UNSET;
ret = mpath_prin_activepath(mpp, MPATH_PRIN_RKEY_SA, resp, noisy);
@@ -928,15 +928,15 @@ int update_map_pr(struct multipath *mpp)
if (ret != MPATH_PR_SUCCESS )
{
condlog(0,"%s : pr in read keys service action failed Error=%d", mpp->alias, ret);
- free(resp);
- return ret;
+ goto out;
}
+ ret = MPATH_PR_SUCCESS;
+
if (resp->prin_descriptor.prin_readkeys.additional_length == 0 )
{
condlog(3,"%s: No key found. Device may not be registered. ", mpp->alias);
- free(resp);
- return MPATH_PR_SUCCESS;
+ goto out;
}
condlog(2, "%s: Multipath reservation_key: 0x%" PRIx64 " ", mpp->alias,
@@ -961,6 +961,7 @@ int update_map_pr(struct multipath *mpp)
condlog(2, "%s: prflag flag set.", mpp->alias );
}
+out:
free(resp);
- return MPATH_PR_SUCCESS;
+ return ret;
}

View File

@ -0,0 +1,300 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Muneendra <muneendra.kumar@broadcom.com>
Date: Wed, 20 Sep 2023 20:41:15 -0700
Subject: [PATCH] multipathd: Added support to handle FPIN-Li events for
FC-NVMe
This patch adds the support to handle FPIN-Li for FC-NVMe.
On receiving the FPIN-Li events this patch moves the devices paths
which are affected due to link integrity to marginal path groups.
The paths which are set to marginal path group will be unset
on receiving the RSCN events
(mwilck: minor compile fix for 32-bit architectures)
Signed-off-by: Muneendra <muneendra.kumar@broadcom.com>
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
---
multipathd/fpin_handlers.c | 206 +++++++++++++++++++++++++++----------
1 file changed, 151 insertions(+), 55 deletions(-)
diff --git a/multipathd/fpin_handlers.c b/multipathd/fpin_handlers.c
index 599f2893..6b90cded 100644
--- a/multipathd/fpin_handlers.c
+++ b/multipathd/fpin_handlers.c
@@ -60,18 +60,15 @@ static void _udev_device_unref(void *p)
/*set/unset the path state to marginal*/
-static int fpin_set_pathstate(struct path *pp, bool set)
+static void fpin_set_pathstate(struct path *pp, bool set)
{
const char *action = set ? "set" : "unset";
- if (!pp || !pp->mpp || !pp->mpp->alias)
- return -1;
-
- condlog(3, "\n%s: %s marginal path %s (fpin)",
- action, pp->mpp->alias, pp->dev_t);
+ condlog(3, "%s: %s marginal path %s (fpin)",
+ pp->mpp ? pp->mpp->alias : "orphan", action, pp->dev_t);
pp->marginal = set;
- pp->mpp->fpin_must_reload = true;
- return 0;
+ if (pp->mpp)
+ pp->mpp->fpin_must_reload = true;
}
/* This will unset marginal state of a device*/
@@ -82,14 +79,14 @@ static void fpin_path_unsetmarginal(char *devname, struct vectors *vecs)
pp = find_path_by_dev(vecs->pathvec, devname);
if (!pp)
pp = find_path_by_devt(vecs->pathvec, devname);
-
- fpin_set_pathstate(pp, false);
+ if (pp)
+ fpin_set_pathstate(pp, false);
}
/*This will set the marginal state of a device*/
-static int fpin_path_setmarginal(struct path *pp)
+static void fpin_path_setmarginal(struct path *pp)
{
- return fpin_set_pathstate(pp, true);
+ fpin_set_pathstate(pp, true);
}
/* Unsets all the devices in the list from marginal state */
@@ -176,8 +173,8 @@ static void fpin_set_rport_marginal(struct udev_device *rport_dev)
"Marginal", strlen("Marginal"));
}
-/*Add the marginal devices info into the list*/
-static void
+/*Add the marginal devices info into the list and return 0 on success*/
+static int
fpin_add_marginal_dev_info(uint32_t host_num, char *devname)
{
struct marginal_dev_list *newdev = NULL;
@@ -192,65 +189,160 @@ fpin_add_marginal_dev_info(uint32_t host_num, char *devname)
list_add_tail(&(newdev->node),
&fpin_li_marginal_dev_list_head);
pthread_mutex_unlock(&fpin_li_marginal_dev_mutex);
- }
+ } else
+ return -ENOMEM;
+ return 0;
}
/*
- * This function goes through the vecs->pathvec, and for
- * each path, check that the host number,
- * the target WWPN associated with the path matches
- * with the els wwpn and sets the path and port state to
+ * This function compares Transport Address Controller Port pn,
+ * Host Transport Address Controller Port pn with the els wwpn ,attached_wwpn
+ * and return 1 (match) or 0 (no match) or a negative error code
+ */
+static int extract_nvme_addresses_chk_path_pwwn(const char *address,
+ uint64_t els_wwpn, uint64_t els_attached_wwpn)
+
+{
+ uint64_t traddr;
+ uint64_t host_traddr;
+
+ /*
+ * Find the position of "traddr=" and "host_traddr="
+ * and the address will be in the below format
+ * "traddr=nn-0x200400110dff9400:pn-0x200400110dff9400,
+ * host_traddr=nn-0x200400110dff9400:pn-0x200400110dff9400"
+ */
+ const char *traddr_start = strstr(address, "traddr=");
+ const char *host_traddr_start = strstr(address, "host_traddr=");
+
+ if (!traddr_start || !host_traddr_start)
+ return -EINVAL;
+
+ /* Extract traddr pn */
+ if (sscanf(traddr_start, "traddr=nn-%*[^:]:pn-%" SCNx64, &traddr) != 1)
+ return -EINVAL;
+
+ /* Extract host_traddr pn*/
+ if (sscanf(host_traddr_start, "host_traddr=nn-%*[^:]:pn-%" SCNx64,
+ &host_traddr) != 1)
+ return -EINVAL;
+ condlog(4, "traddr 0x%" PRIx64 " hosttraddr 0x%" PRIx64 " els_wwpn 0x%"
+ PRIx64" els_host_traddr 0x%" PRIx64,
+ traddr, host_traddr,
+ els_wwpn, els_attached_wwpn);
+ if ((host_traddr == els_attached_wwpn) && (traddr == els_wwpn))
+ return 1;
+ return 0;
+}
+
+/*
+ * This function check that the Transport Address Controller Port pn,
+ * Host Transport Address Controller Port pn associated with the path matches
+ * with the els wwpn ,attached_wwpn and sets the path state to
* Marginal
*/
-static int fpin_chk_wwn_setpath_marginal(uint16_t host_num, struct vectors *vecs,
+static void fpin_check_set_nvme_path_marginal(uint16_t host_num, struct path *pp,
+ uint64_t els_wwpn, uint64_t attached_wwpn)
+{
+ struct udev_device *ctl = NULL;
+ const char *address = NULL;
+ int ret = 0;
+
+ ctl = udev_device_get_parent_with_subsystem_devtype(pp->udev, "nvme", NULL);
+ if (ctl == NULL) {
+ condlog(2, "%s: No parent device for ", pp->dev);
+ return;
+ }
+ address = udev_device_get_sysattr_value(ctl, "address");
+ if (!address) {
+ condlog(2, "%s: unable to get the address ", pp->dev);
+ return;
+ }
+ condlog(4, "\n address %s: dev :%s\n", address, pp->dev);
+ ret = extract_nvme_addresses_chk_path_pwwn(address, els_wwpn, attached_wwpn);
+ if (ret <= 0)
+ return;
+ ret = fpin_add_marginal_dev_info(host_num, pp->dev);
+ if (ret < 0)
+ return;
+ fpin_path_setmarginal(pp);
+}
+
+/*
+ * This function check the host number, the target WWPN
+ * associated with the path matches with the els wwpn and
+ * sets the path and port state to Marginal
+ */
+static void fpin_check_set_scsi_path_marginal(uint16_t host_num, struct path *pp,
uint64_t els_wwpn)
{
- struct path *pp;
- struct multipath *mpp;
- int i, k;
char rport_id[42];
const char *value = NULL;
struct udev_device *rport_dev = NULL;
uint64_t wwpn;
int ret = 0;
+ sprintf(rport_id, "rport-%d:%d-%d",
+ pp->sg_id.host_no, pp->sg_id.channel, pp->sg_id.transport_id);
+ rport_dev = udev_device_new_from_subsystem_sysname(udev,
+ "fc_remote_ports", rport_id);
+ if (!rport_dev) {
+ condlog(2, "%s: No fc_remote_port device for '%s'", pp->dev,
+ rport_id);
+ return;
+ }
+ pthread_cleanup_push(_udev_device_unref, rport_dev);
+ value = udev_device_get_sysattr_value(rport_dev, "port_name");
+ if (!value)
+ goto unref;
+
+ wwpn = strtol(value, NULL, 16);
+ /*
+ * If the port wwpn matches sets the path and port state
+ * to marginal
+ */
+ if (wwpn == els_wwpn) {
+ ret = fpin_add_marginal_dev_info(host_num, pp->dev);
+ if (ret < 0)
+ goto unref;
+ fpin_path_setmarginal(pp);
+ fpin_set_rport_marginal(rport_dev);
+ }
+unref:
+ pthread_cleanup_pop(1);
+ return;
+
+}
+
+/*
+ * This function goes through the vecs->pathvec, and for
+ * each path, it checks and sets the path state to marginal
+ * if the path's associated port wwpn ,hostnum matches with
+ * els wwnpn ,attached_wwpn
+ */
+static int fpin_chk_wwn_setpath_marginal(uint16_t host_num, struct vectors *vecs,
+ uint64_t els_wwpn, uint64_t attached_wwpn)
+{
+ struct path *pp;
+ struct multipath *mpp;
+ int i, k;
+ int ret = 0;
pthread_cleanup_push(cleanup_lock, &vecs->lock);
lock(&vecs->lock);
pthread_testcancel();
vector_foreach_slot(vecs->pathvec, pp, k) {
- /* Checks the host number and also for the SCSI FCP */
- if (pp->bus != SYSFS_BUS_SCSI || pp->sg_id.proto_id != SCSI_PROTOCOL_FCP || host_num != pp->sg_id.host_no)
+ if (!pp->mpp)
continue;
- sprintf(rport_id, "rport-%d:%d-%d",
- pp->sg_id.host_no, pp->sg_id.channel, pp->sg_id.transport_id);
- rport_dev = udev_device_new_from_subsystem_sysname(udev,
- "fc_remote_ports", rport_id);
- if (!rport_dev) {
- condlog(2, "%s: No fc_remote_port device for '%s'", pp->dev,
- rport_id);
- continue;
- }
- pthread_cleanup_push(_udev_device_unref, rport_dev);
- value = udev_device_get_sysattr_value(rport_dev, "port_name");
- if (!value)
- goto unref;
-
- if (value)
- wwpn = strtol(value, NULL, 16);
- /*
- * If the port wwpn matches sets the path and port state
- * to marginal
- */
- if (wwpn == els_wwpn) {
- ret = fpin_path_setmarginal(pp);
- if (ret < 0)
- goto unref;
- fpin_set_rport_marginal(rport_dev);
- fpin_add_marginal_dev_info(host_num, pp->dev);
+ /*checks if the bus type is nvme and the protocol is FC-NVMe*/
+ if ((pp->bus == SYSFS_BUS_NVME) && (pp->sg_id.proto_id == NVME_PROTOCOL_FC)) {
+ fpin_check_set_nvme_path_marginal(host_num, pp, els_wwpn, attached_wwpn);
+ } else if ((pp->bus == SYSFS_BUS_SCSI) &&
+ (pp->sg_id.proto_id == SCSI_PROTOCOL_FCP) &&
+ (host_num == pp->sg_id.host_no)) {
+ /* Checks the host number and also for the SCSI FCP */
+ fpin_check_set_scsi_path_marginal(host_num, pp, els_wwpn);
}
-unref:
- pthread_cleanup_pop(1);
}
/* walk backwards because update_path_groups() can remove mpp */
vector_foreach_slot_backwards(vecs->mpvec, mpp, i) {
@@ -279,14 +371,18 @@ fpin_parse_li_els_setpath_marginal(uint16_t host_num, struct fc_tlv_desc *tlv,
struct fc_fn_li_desc *li_desc = (struct fc_fn_li_desc *)tlv;
int count = 0;
int ret = 0;
+ uint64_t attached_wwpn;
/* Update the wwn to list */
wwn_count = be32_to_cpu(li_desc->pname_count);
- condlog(4, "Got wwn count as %d\n", wwn_count);
+ attached_wwpn = be64_to_cpu(li_desc->attached_wwpn);
+ condlog(4, "Got wwn count as %d detecting wwn 0x%" PRIx64
+ " attached_wwpn 0x%" PRIx64 "\n",
+ wwn_count, be64_to_cpu(li_desc->detecting_wwpn), attached_wwpn);
for (iter = 0; iter < wwn_count; iter++) {
wwpn = be64_to_cpu(li_desc->pname_list[iter]);
- ret = fpin_chk_wwn_setpath_marginal(host_num, vecs, wwpn);
+ ret = fpin_chk_wwn_setpath_marginal(host_num, vecs, wwpn, attached_wwpn);
if (ret < 0)
condlog(2, "failed to set the path marginal associated with wwpn: 0x%" PRIx64 "\n", wwpn);

View File

@ -0,0 +1,76 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Benjamin Marzinski <bmarzins@redhat.com>
Date: Wed, 22 Nov 2023 16:41:22 -0500
Subject: [PATCH] multipathd: Make sure to disable queueing if recovery has
failed.
If a multipath device has no_path_retry set to a number and has lost all
paths, gone into recovery mode, and timed out, it will disable
queue_if_no_paths. After that, if the device is reloaded by multipath
outside of multipathd, it will re-enable queuieng on the device. When
multipathd later calls set_no_path_retry() to update the queueing state,
it will not disable queue_if_no_paths, since the device is still in the
recovery state, so it believes no work needs to be done. The device will
remain in the recovery state, with retry_ticks at 0, and queueing
enabled, even though there are no usable paths.
To fix this, in set_no_path_retry(), if no_path_retry is set to a number
and the device is queueing but it is in recovery mode and out of
retries with no usable paths, manually disable queue_if_no_path.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
libmultipath/structs_vec.c | 26 ++++++++++++++++++++++++--
1 file changed, 24 insertions(+), 2 deletions(-)
diff --git a/libmultipath/structs_vec.c b/libmultipath/structs_vec.c
index 9613252f..0304777b 100644
--- a/libmultipath/structs_vec.c
+++ b/libmultipath/structs_vec.c
@@ -377,8 +377,19 @@ void __set_no_path_retry(struct multipath *mpp, bool check_features)
!mpp->in_recovery)
dm_queue_if_no_path(mpp->alias, 1);
leave_recovery_mode(mpp);
- } else
- enter_recovery_mode(mpp);
+ } else {
+ /*
+ * If in_recovery is set, enter_recovery_mode does
+ * nothing. If the device is already in recovery
+ * mode and has already timed out, manually call
+ * dm_queue_if_no_path to stop it from queueing.
+ */
+ if ((!mpp->features || is_queueing) &&
+ mpp->in_recovery && mpp->retry_tick == 0)
+ dm_queue_if_no_path(mpp->alias, 0);
+ if (pathcount(mpp, PATH_PENDING) == 0)
+ enter_recovery_mode(mpp);
+ }
break;
}
}
@@ -521,6 +532,11 @@ int verify_paths(struct multipath *mpp, struct vectors *vecs)
* -1 (FAIL) : fail_if_no_path
* 0 (UNDEF) : nothing
* >0 : queue_if_no_path enabled, turned off after polling n times
+ *
+ * Since this will only be called when fail_path(), update_multipath(), or
+ * io_err_stat_handle_pathfail() are failing a previously active path, the
+ * device cannot already be in recovery mode, so there will never be a need
+ * to disable queueing here.
*/
void update_queue_mode_del_path(struct multipath *mpp)
{
@@ -534,6 +550,12 @@ void update_queue_mode_del_path(struct multipath *mpp)
condlog(2, "%s: remaining active paths: %d", mpp->alias, active);
}
+/*
+ * Since this will only be called from check_path() -> reinstate_path() after
+ * the queueing state has been updated in set_no_path_retry, this does not
+ * need to worry about modifying the queueing state except when actually
+ * leaving recovery mode.
+ */
void update_queue_mode_add_path(struct multipath *mpp)
{
int active = count_active_paths(mpp);

View File

@ -1,7 +1,7 @@
Summary: Tools to manage multipath devices using device-mapper
Name: device-mapper-multipath
Version: 0.8.4
Release: 22%{?dist}
Release: 41%{?dist}
License: GPLv2
Group: System Environment/Base
URL: http://christophe.varoqui.free.fr/
@ -101,12 +101,60 @@ Patch00087: 0087-multipathd-avoid-unnecessary-path-read-only-reloads.patch
Patch00088: 0088-libmultipath-make-helper-function-to-trigger-path-ue.patch
Patch00089: 0089-multipathd-trigger-udev-change-on-path-addition.patch
Patch00090: 0090-RH-add-support-to-mpathconf-for-setting-arbitrary-de.patch
Patch00091: 0091-multipath-tools-add-HPE-MSA-1060-2060-to-hwtable.patch
Patch00092: 0092-multipath-tools-update-mpp-force_readonly-in-ev_add_.patch
Patch00093: 0093-updated-HPE-MSA-builtin-config.patch
Patch00094: 0094-multipath-return-failure-on-an-invalid-remove-comman.patch
Patch00095: 0095-libmultipath-steal-the-src-string-pointer-in-merge_s.patch
Patch00096: 0096-libmultipath-enable-linear-ordering-of-bus-proto-tup.patch
Patch00097: 0097-libmultipath-use-bus_protocol_id-in-snprint_path_pro.patch
Patch00098: 0098-libmultipath-make-protocol_name-global.patch
Patch00099: 0099-libmultipath-add-a-protocol-subsection-to-multipath..patch
Patch00100: 0100-libmultipath-Set-the-scsi-timeout-parameters-by-path.patch
Patch00101: 0101-libmultipath-check-the-overrides-pctable-for-path-va.patch
Patch00102: 0102-libmultipath-fix-eh_deadline-documentation.patch
Patch00103: 0103-libmultipath-Add-documentation-for-the-protocol-subs.patch
Patch00104: 0104-libmultipath-use-symbolic-value-for-invalid-pcentry.patch
Patch00105: 0105-multipathd-handle-fpin-events.patch
Patch00106: 0106-multipathd-disallow-changing-to-from-fpin-marginal-p.patch
Patch00107: 0107-libmultipath-unset-detect_checker-for-clariion-Unity.patch
Patch00108: 0108-multipathd-Add-missing-ctype-include.patch
Patch00109: 0109-multipathd-replace-libreadline-with-libedit.patch
Patch00110: 0110-multipath-fix-systemd-timers-in-the-initramfs.patch
Patch00111: 0111-multipathd-factor-out-the-code-to-flush-a-map-with-n.patch
Patch00112: 0112-libmultipath-return-success-if-we-raced-to-remove-a-.patch
Patch00113: 0113-multipathd-Handle-losing-all-path-in-update_map.patch
Patch00114: 0114-multipathd-ignore-duplicated-multipathd-command-keys.patch
Patch00115: 0115-multipath-tools-use-run-instead-of-dev-shm.patch
Patch00116: 0116-kpartx-hold-device-open-until-partitions-have-been-c.patch
Patch00117: 0117-libmultipath-cleanup-remove_feature.patch
Patch00118: 0118-libmultipath-cleanup-add_feature.patch
Patch00119: 0119-multipath-tests-tests-for-adding-and-removing-featur.patch
Patch00120: 0120-libmultipath-fix-queue_mode-feature-handling.patch
Patch00121: 0121-multipath-tests-tests-for-reconcile_features_with_qu.patch
Patch00122: 0122-libmultipath-prepare-proto_id-for-use-by-non-scsi-de.patch
Patch00123: 0123-libmultipath-get-nvme-path-transport-protocol.patch
Patch00124: 0124-libmultipath-enforce-queue_mode-bio-for-nmve-tcp-pat.patch
Patch00125: 0125-multipath-add-historical-service-time-to-the-man-pag.patch
Patch00126: 0126-libmultipath-copy-mpp-hwe-from-pp-hwe.patch
Patch00127: 0127-libmultipath-don-t-leak-memory-on-invalid-strings.patch
Patch00128: 0128-libmutipath-validate-the-argument-count-of-config-st.patch
Patch00129: 0129-libmultipath-select-resize-action-even-if-reload-is-.patch
Patch00130: 0130-libmultipath-cleanup-ACT_CREATE-code-in-select_actio.patch
Patch00131: 0131-libmultipath-keep-renames-from-stopping-other-multip.patch
Patch00132: 0132-multipathd-make-pr-registration-consistent.patch
Patch00133: 0133-libmultipath-make-prflag-an-enum.patch
Patch00134: 0134-multipathd-handle-no-active-paths-in-update_map_pr.patch
Patch00135: 0135-libmpathpersist-fix-resource-leak-in-update_map_pr.patch
Patch00136: 0136-multipathd-Added-support-to-handle-FPIN-Li-events-fo.patch
Patch00137: 0137-multipathd-Make-sure-to-disable-queueing-if-recovery.patch
# runtime
Requires: %{name}-libs = %{version}-%{release}
Requires: kpartx = %{version}-%{release}
Requires: device-mapper >= 1.02.96
Requires: userspace-rcu
Requires: libedit
Requires(post): systemd-units
Requires(preun): systemd-units
Requires(postun): systemd-units
@ -114,7 +162,7 @@ Requires(postun): systemd-units
# build/setup
BuildRequires: libaio-devel, device-mapper-devel >= 1.02.89
BuildRequires: libselinux-devel, libsepol-devel
BuildRequires: readline-devel, ncurses-devel
BuildRequires: libedit-devel, ncurses-devel
BuildRequires: systemd-units, systemd-devel
BuildRequires: json-c-devel, perl-interpreter, pkgconfig, gcc
BuildRequires: userspace-rcu-devel
@ -198,7 +246,8 @@ make install \
rcdir=%{_initrddir} \
unitdir=%{_unitdir} \
includedir=%{_includedir} \
pkgconfdir=%{_pkgconfdir}
pkgconfdir=%{_pkgconfdir} \
tmpfilesdir=%{_tmpfilesdir}
# tree fix up
install -d %{buildroot}/etc/multipath
@ -237,6 +286,7 @@ fi
%{_mandir}/man8/mpathpersist.8.gz
%config %{_udevrulesdir}/62-multipath.rules
%config %{_udevrulesdir}/11-dm-mpath.rules
%{_tmpfilesdir}/multipath.conf
%{!?_licensedir:%global license %%doc}
%license LICENSES/GPL-2.0 LICENSES/LGPL-2.0
%doc README
@ -276,7 +326,7 @@ fi
%{!?_licensedir:%global license %%doc}
%license LICENSES/GPL-2.0
%{_sbindir}/kpartx
/usr/lib/udev/kpartx_id
%{_udevrulesdir}/../kpartx_id
%{_mandir}/man8/kpartx.8.gz
%config %{_udevrulesdir}/11-dm-parts.rules
%config %{_udevrulesdir}/66-kpartx.rules
@ -302,6 +352,132 @@ fi
%{_pkgconfdir}/libdmmp.pc
%changelog
* Fri Jan 5 2024 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-41
- Add 0137-multipathd-Make-sure-to-disable-queueing-if-recovery.patch
- Resolves: RHEL-16563
* Thu Nov 2 2023 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-40
- Add 0136-multipathd-Added-support-to-handle-FPIN-Li-events-fo.patch
* Add support in multipathd for NVMe to listen for FPIN-Li events and
mark effected paths as marginal
- Resolves: RHEL-6677
* Thu Mar 16 2023 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-39
- Add OSCI tests directory
- Make kpartx_id installation location relative to %{_udevrulesdir}
- Resolves: bz #2164871
* Wed Mar 15 2023 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-38
- Add 0132-multipathd-make-pr-registration-consistent.patch
- Add 0133-libmultipath-make-prflag-an-enum.patch
- Add 0134-multipathd-handle-no-active-paths-in-update_map_pr.patch
- Add 0135-libmpathpersist-fix-resource-leak-in-update_map_pr.patch
- Resolves: bz #2164871
* Fri Feb 3 2023 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-37
- Fix bugzilla linked to the changes (was previously linked to
the wrong bug, 2162537)
- Resolves: bz #2166468
* Wed Feb 1 2023 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-36
- Add 0129-libmultipath-select-resize-action-even-if-reload-is-.patch
- Add 0130-libmultipath-cleanup-ACT_CREATE-code-in-select_actio.patch
- Add 0131-libmultipath-keep-renames-from-stopping-other-multip.patch
- Resolves: bz #2166468
* Mon Jan 16 2023 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-35
- Add 0127-libmultipath-don-t-leak-memory-on-invalid-strings.patch
- Add 0128-libmutipath-validate-the-argument-count-of-config-st.patch
- Resolves: bz #2155560
* Tue Nov 29 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-34
- Add 0126-libmultipath-copy-mpp-hwe-from-pp-hwe.patch
* Fixes bz #2126714
- Cleanup multiple CI tests
- Resolves: bz #2126714
* Wed Nov 23 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-33
- Add 0125-multipath-add-historical-service-time-to-the-man-pag.patch
* Fixes bz #2141996
- Modify tests/multipath_conf_syntax/main.sh
* fix unrelated test error
- Resolves: bz #2141996
* Thu Nov 10 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-32
- Add 0116-kpartx-hold-device-open-until-partitions-have-been-c.patch
* Fixes bz #2128885
- Add 0117-libmultipath-cleanup-remove_feature.patch
- Add 0118-libmultipath-cleanup-add_feature.patch
- Add 0119-multipath-tests-tests-for-adding-and-removing-featur.patch
- Add 0120-libmultipath-fix-queue_mode-feature-handling.patch
- Add 0121-multipath-tests-tests-for-reconcile_features_with_qu.patch
- Add 0122-libmultipath-prepare-proto_id-for-use-by-non-scsi-de.patch
- Add 0123-libmultipath-get-nvme-path-transport-protocol.patch
- Add 0124-libmultipath-enforce-queue_mode-bio-for-nmve-tcp-pat.patch
* Fixes bz #2022359
- Resolves: bz #2022359, #2128885
* Thu Oct 13 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-31
- Add 0114-multipathd-ignore-duplicated-multipathd-command-keys.patch
* Fixes bz #2133996
- Add 0115-multipath-tools-use-run-instead-of-dev-shm.patch
* Fixes bz #2133990
- Resolves: bz #2133990, #2133996
* Fri Sep 9 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-30
- Add 0111-multipathd-factor-out-the-code-to-flush-a-map-with-n.patch
- Add 0112-libmultipath-return-success-if-we-raced-to-remove-a-.patch
- Add 0113-multipathd-Handle-losing-all-path-in-update_map.patch
- Resolves: bz #2110485
* Fri Sep 2 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-29
- Rebuild for rhel-8.8.0
- Resolves: bz #2123446
* Wed Aug 24 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-28
- Add 0110-multipath-fix-systemd-timers-in-the-initramfs.patch
- Resolves: bz #1916168
* Fri Aug 19 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-27
- Add 0108-multipathd-Add-missing-ctype-include.patch
- Add 0109-multipathd-replace-libreadline-with-libedit.patch
* replace readline with libedit, to avoid license conflicts
- Resolves: bz #2119887
* Wed Jun 8 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-26
- Add 0107-libmultipath-unset-detect_checker-for-clariion-Unity.patch
- Resolves: bz #2082205
* Tue May 17 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-25
- Add 0105-multipathd-handle-fpin-events.patch
- Add 0106-multipathd-disallow-changing-to-from-fpin-marginal-p.patch
- Resolves: bz #2083077
* Mon May 16 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-24
- Add 0094-multipath-return-failure-on-an-invalid-remove-comman.patch
* Fixes bz #2052054
- Add 0095-libmultipath-steal-the-src-string-pointer-in-merge_s.patch
- Add 0096-libmultipath-enable-linear-ordering-of-bus-proto-tup.patch
- Add 0097-libmultipath-use-bus_protocol_id-in-snprint_path_pro.patch
- Add 0098-libmultipath-make-protocol_name-global.patch
- Add 0099-libmultipath-add-a-protocol-subsection-to-multipath..patch
- Add 0100-libmultipath-Set-the-scsi-timeout-parameters-by-path.patch
- Add 0101-libmultipath-check-the-overrides-pctable-for-path-va.patch
- Add 0102-libmultipath-fix-eh_deadline-documentation.patch
- Add 0103-libmultipath-Add-documentation-for-the-protocol-subs.patch
- Add 0104-libmultipath-use-symbolic-value-for-invalid-pcentry.patch
* The above 10 patches implement feature from bz #2065477
- Resolves: bz #2052054, #2065477
* Tue May 10 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-23
- Add 0091-multipath-tools-add-HPE-MSA-1060-2060-to-hwtable.patch
* Partial fix for bz #2058222
- Add 0092-multipath-tools-update-mpp-force_readonly-in-ev_add_.patch
* Fixes bz #2065468
- Add 0093-updated-HPE-MSA-builtin-config.patch
* Fixes bz #2058222
- Resolves bz #2065468, #2058222
* Mon Feb 7 2022 Benjamin Marzinski <bmarzins@redhat.com> 0.8.4-22
- Add 0090-RH-add-support-to-mpathconf-for-setting-arbitrary-de.patch
- Resolves: bz #1981240