With multiple kernel variants on the same architecture, e.g. the 4k and
64k kernel on aarch64, we can no longer assume that the crashkernel
value for the currently running kernel will work for all installed
kernels. This also means that we can no longer update the grub config as
we don't know which value to set it to. Thus get the crashkernel value
for each kernel and stop updating the grub config.
While at it merge the _new_fadump and _fadump_val variables and remove
_read_kernel_arg_in_grub_etc_default which has no user.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
Prevent the --fadump option to be used on non-PPC systems. This not only
prevents user errors but also guarantees that _dump_mode and _fadump_val are
empty on these systems.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
_update_kernel_cmdline handles two cmdline parameters at once. This does not
only make the function itself but also its callers more complicated than
necessary. For example in _update_crashkernel the fadump gets "updated" to
the value that has been read from grubby. Thus simplify
_update_kernel_cmdline to only update one parameter at once.
While at it shorten some variable named in the callers.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
condrestart is a left over from the time of SysVinit that is no longer
needed since the kexec-tools switched to systemd (10c91a1 ("Removing
sysvinit files") plus the one before). What's especially intriguing is
that from the beginning (0112f36 ("- Add a kdump sysconfig file and init
script - Spec file additions for pre/post install/uninstall")) the
sub-command never did any actual work (other than not returning an
error). Thus simply remove the condrestart sub-command.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
_get_current_running_kernel_path is identical to
_find_kernel_path_by_release $(uname -r) so simply use this instead of
defining a new function.
While at it simplify reset_crashkernel slightly. This changes the
behavior of the function for the case when KDUMP_KERNELVER is defined
but no kernel with this version is installed. Before, the missing
kernel is silently ignored and the currently running kernel is used
instead. Now, kdumpctl will exit with an error.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
Only check whether modules for a given kernel version are installed
instead of searching for a kernel image. It's safer to assume that every
kernel uses kernel modules compared to that it follows certain naming
conventions. Furthermore it is much more lightweight and thus allows to
determine the KDUMP_KERNELVER much earlier for every command in
kdumpctl.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
There is value to use the function in other places as well. For example
it can be used to check whether optional dependencies, like grubby, are
installed. Thus make it more generic so it can be reused in later
commits.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
This includes fixes for
SC2295 (info): Expansions inside ${..} need to be quoted separately, otherwise they match as patterns.
SC2005 (style): Useless echo? Instead of 'echo $(cmd)', just use 'cmd'.
SC2162 (info): read without -r will mangle backslashes.
SC2086 (info): Double quote to prevent globbing and word splitting.
SC2317 (info): Command appears to be unreachable. Check usage (or ignore if invoked indirectly).
In addition add some source hints to prevent false positive findings.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
The fadump sysfs nodes /sys/kernel/fadump_[enabled|registered], have
been relocated to /sys/kernel/fadump/[enabled|registered] by kernel
commits d418b19f34ed ("powerpc/fadump: Reorganize /sys/kernel/fadump_*
sysfs files").
To ensure compatibility, symbolic links were added for each relocated
sysfs entry. Nonetheless, note that these symbolic links might be
removed later, as they have been deprecated by kernel commit
3f5f1f22ef10 ("Documentation/ABI: Mark /sys/kernel/fadump_* sysfs files
deprecated")
This patch updates the scripts to use the updated fadump sysfs files.
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
The size of the reserved memory in the functions show_reserved_mem,
check_crash_mem_reserved, and do_estimate are fetched from the sysfs
node `/sys/kernel/kexec_crash_size`. However, in the case of fadump,
the reserved area size is instead present in
/sys/kernel/fadump/mem_reserved.
For example:
$ kdumpctl showmem
kdump: Dump mode is fadump
kdump: Reserved 0MB memory for crash kernel
The above command showed 0MB for Reserved memory which is incorrect, the
actual reservation was 2048MB.
To resolve this issue a new helper function is introduced to fetch
reserved memory size based on the dump mode. For "fadump" mode,
it looks in `/sys/kernel/fadump/mem_reserved`, otherwise, it uses
`/sys/kernel/kexec_crash_size`. And all functions that previously
fetching reserved memory directly from `/sys/kernel/kexec_crash_size`
sysfs node are now updated to use this new function to get the reserved
memory size.
With the fix in place, the `kdumpctl showmem` command will now display
correct reserved memory size.
$ kdumpctl showmem
kdump: Dump mode is fadump
kdump: Reserved 2048MB memory for crash kernel
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Reported-by: Sachin P Bappalige <sachinpb@linux.vnet.ibm.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
There is confusing message in /var/log/anaconda/packaging.log when installing
kexec-tools during the system installation on ppc64le:
Event Notification Registration successful (id: 1)
Make servicelog_notify slient when there are no erros.
Signed-off-by: Lichen Liu <lichliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
For kernel 64k variant, it terminates with substring 64k-debug, e.g.
vmlinuz-5.14.0-327.el9.aarch64+64k-debug.
Providing an extra matching pattern to filter out it.
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The temporary directory is currently created under the current working
directory. That alone isn't ideal but works most of the time. However,
it will fail when the current working directory is not writable. So make
sure the directory is created within TMPDIR.
Fixes: ea00b7d ("kdumpctl: Move temp file in get_kernel_size to global temp dir")
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
And with commit t5b31b099 ("Simplify the management of the kernel
parameter crashkernel"), the default crashkernel value will be
used for the kernel. But the test VM has a RAM of 768M thus this is no
actual reserved memory for kdump. Even With the old crashkernel=224M,
network dumping tests like nfs-kdump will fail out of memory when
running against current Fedora Cloud images (>=F37).
This patch address the above two issues by
1. increasing the RAM of test VM to 1G
2. installing the kernel-modules which contains the squashfs module in
order to use the dracut squash module for kdump initrd.
Thanks to the dracut squash module, now even crashkernel=192M (the
default crashkernel value for RAM between 1G and 4G) works for
network dumping. Another benefit brought by this change is the default
crashkernel value can be tested as well.
Signed-off-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
All the tests failed to run on the Fedora 37 host because the boot
partition failed to be mounted and in turn the key kernel cmdline
parameters like selinux=0 couldn't be added.
The root problem is somehow running lsblk on the second partition
returns an empty label unless we wait for enough time. Before figuring
out the root cause, simply skip check that the second partition
needs to have the boot label.
Note the root problem can be produced by building a test image,
cd tests
./scripts/build-image.sh Fedora-Cloud-Base-37-1.7.x86_64.qcow2 output_image scripts/build-scripts/base-image_test.sh
Source image is qcow2, using snapshot...
Formatting 'build/base-image1.building', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=5368709120 backing_file=Fedora-Cloud-Base-37-1.7.x86_64.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
It's a image with multiple partitions, using last partition as main partition
grep: /boot/grub2/grubenv: No such file or directory
grub2-editenv: error: cannot open `/boot/grub2/grubenv.new': No such file or directory.
/dev/nbd0 disconnected
Signed-off-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Tests failed to run against Fedora 37 or newer cloud images because of
the following error,
It's a image with multiple partitions, using last partition as main partition
'xxx/tests/build/x86_64/kexec-tools-2.0.26-5.fc37.src.rpm' not found
/dev/nbd0 disconnected
make: *** [Makefile:73: xxx/tests/output/test-base-image] Error 1
This is because starting with Fedora 37, rpm changes its API,
# Fedora >= 37
$ rpm -q --specfile kexec-tools.spec
kexec-tools-2.0.26-5.fc37.src
# Fedora 36
$ rpm -q --specfile kexec-tools.spec
kexec-tools-2.0.26-5.fc36
The tests depends on rpm to generate correct RPM name. Fix this issue by
removing the trailing .src from the output of "rpm -q --specfile".
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Signed-off-by: Coiby Xu <coxu@redhat.com>
After introducing 64k variant kernel on aarch64, an example kernel name
looks like "vmlinuz-5.14.0-316.el9.aarch64+64k". To match the plus
symbol, it demands an escape charater.
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
On 4k and 64k kernels, the typical consumption values for SMMU are 36MB
and 384MB, respectively. Hence for 64k kernel, the consumption by smmu
should be taken into account carefully.
To do it by adding the extra 384MB value if installing a 64k kernel.
The upper limit value 384MB is calculated according to the formula in
the kernel smmu driver.
As for mlx5 network cards, it is measured by a pratical test, 200M for
64k variant, 150M for 4k variant
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
On aarch64, both 4K and 64K kernel can be installed, while they demand
different size reserved memory for kdump kernel.
'get_conf PAGE_SIZE' can not work if installing a 64K kernel when
running a 4K kernel. Hence resorting to the kernel release naming rules.
At present, the 64K kernel has the keyword '64k' in its suffix.
The base line for 64K is decided based on 4K. The diff 100M is picked up
since on a high end machine without smmu enabled, the diff of MemFree is
82M.
As for the smmu case, a huge difference in the memory consumption lies
between 64k and 4k driver. And it should be calculated separatedly.
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
kdump_get_arch_recommend_crashkernel() expects the kernel version info,
while _update_kernel() provides the absolute path, which contains the
kernel version info.
This patch introduce a dedicated function parse_kver_from_path() to
extract the kernel info from the path
Credit to Philipp, who contributes the original code.
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
This help function can manipulate the crashkernel cmdline by adding an
number for each item. Also a basic test case for _crashkernel_add() is
provided in this patch.
Credit to Philipp, who contributes the original code.
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
The warning messages should not be included in the generated files.
Redirecting the warning for an unknown architecture to stderr.
Signed-off-by: Lichen Liu <lichliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
A space was added by mistake and unfortunately fips-mode-setup refuses
an extra parameter,
# fips-mode-setup --is-enabled 2 > /dev/null
# echo $?
2
# fips-mode-setup --is-enabled 2
Check, enable, or disable the system FIPS mode.
usage: /usr/bin/fips-mode-setup --enable|--disable [--no-bootcfg]
usage: /usr/bin/fips-mode-setup --check
usage: /usr/bin/fips-mode-setup --is-enabled
So in this case mkdumprd can never detect if FIPS is enabled. Fix this
mistake.
Fixes: 443a43e0 ("mkdumprd: call dracut with --add-device to install the drivers needed by /boot partition automatically for FIPS")
Signed-off-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Tao Liu <ltao@redhat.com>
Currently, kexec-tools only updates the crashkernel to a new default
value only when both two conditions are met,
- auto_reset_crashkernel=yes in kdump.conf
- existing kernels or current running kernel should use the old default
value.
To address seen corner cases, the logic to tell if the second condition
is met becomes quite complex. Instead of making the logic more complex
to support aarch64-64k, this patch drops the second condition to
simplify the management of the crashkernel kernel parameter.
Another change brought by this simplification is kexec-tools will also
set up the kernel crashkernel parameter for a fresh install (previously
it's limited to osbuild).
Note
1. This patch also stop trying to update /etc/default/grub because
a) it only affects the static file /boot/grub2/grub.cfg
b) grubby is recommended to change the kernel command-line parameters
for both Fedora [1] and RHEL9 [2][3]
c) For the cases of aarch64 and POWER, different kernels could have
different default crashkernel value.
2. Starting with Fedora 37, posttrans rpm scriplet distinguish between
package install and upgrade.
[1] https://fedoraproject.org/wiki/GRUB_2
[2] https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/9/html/managing_monitoring_and_updating_the_kernel/configuring-kernel-command-line-parameters_managing-monitoring-and-updating-the-kernel#changing-kernel-command-line-parameters-for-all-boot-entries_configuring-kernel-command-line-parameters
[3] https://access.redhat.com/solutions/1136173
Signed-off-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Currently, for non-s390x systems, the return code is 1 even when
_update_kernel_cmdline is correctly executed. This makes callers like
reset_crashkernel_after_update fail to print a message if a kernel has
its crashkernel updated. Fix it by put the code inside if block for
s390x.
Signed-off-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Currently, kdump doesn't work on many FIPS-enabled systems including
Azure, ESXI, Hyper, POWER and etc. When FIPS is enabled, it needs to
access /boot//.vmlinuz-xxx.hmac to verify the integrity of the kernel.
However, on those systems, /boot fails to be mounted due to a lack of
fs and block device drivers and the system just halted after failing to
verify the integrity of the kernel. For example, on Hyper-V, sd_mod, sg,
scsi_transport_fc, hv_storvsc and hv_vmbus need to be installed in order
for /boot to be mounted.
mkdumprd calls dracut with the --no-hostonly-default-device. Following
the documentation (man dracut),
--no-hostonly-default-device
Do not generate implicit host devices like root, swap, fstab, etc.
Use "--mount" or "--add-device" to explicitly add devices as needed
this patch uses "--add-device" to explicitly add the device of /boot.
Note there is already an attempt to fix it in dracut's 01fips module
i.e. via the commit 83651776 ("fips: ensure fs module for /boot is
installed"). Unfortunately it only installs the file system driver e.g.
xfs.
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Signed-off-by: Coiby Xu <coxu@redhat.com>
On aarch64, a 64k kernel's name looks like:
vmlinuz-5.14.0-300.el9.aarch64+64k and the corresponding debug kernel's
name looks like: vmlinuz-5.14.0-300.el9.aarch64+64k-debug, which ends
with the suffix -debug instead of +debug.
Fix the matching pattern by [+|-]debug
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
The default systemd-boot installed kernels on fedora end up in the form:
/boot/efi/36b54597c46383/6.4.0-0.rc0.20230427git6e98b09da931.5.fc39.aarch64/linux
Where the kernel version is a directory containing the kernel (linux)
and the initrd. Thus _find_kernel_path_by release needs to be a bit less
strict and allow some futher characters on the grubby (really bootctl)
output.
Signed-off-by: Jeremy Linton <jeremy.linton@arm.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The new aggressive strip option was added in dracut 058, which tell
dracut to build the initramfs stripping more sections of the ELF
binaries (basically strip .symtab, .strtab).
These section are only useful for debugging runtime failures, but in
kdump kernel, neccessary tools for debug any runtime failure are
absent, there is no point keeping these sections.
Stripping these section can help save some memory with almost no side
effect. So let enable --aggressive-strip by default.
Comparison of unpacked initramfs before / after enabling aggressive strip:
du -hs image image.aggressive-strip
31M image
29M image.aggressive-strip
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
A Unified Kernel Image (UKI) is a single EFI PE executable combining an
EFI stub, a kernel image, an initrd image, and the kernel command line.
They are defined in the Boot Loader Specification [1] as type #2
entries. UKIs have the advantage that all code as well as meta data that
is required to boot the system, not only the kernel image, is combined
in a single PE file and can be signed for EFI SecureBoot. This extends
the coverage of SecureBoot extensively.
For RHEL support for UKI were included into kernel-ark with 16c7e3ee836e
("redhat: Add sub-RPM with a EFI unified kernel image for virtual
machines").
There are two problems with UKIs from the kdump point of view at the
moment. First, they cannot be directly loaded via kexec_file_load and
second, the initrd included isn't suitable for kdump. In order to enable
kdump on systems with UKIs build the kdump initrd as usual and extract
the kernel image before loading the crash kernel.
[1] https://uapi-group.org/specifications/specs/boot_loader_specification/
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
Others will need to use a temporary files, too. In order to avoid
potential clashes of multiple trap handlers move the local temp file
into a global temp dir.
While at it make sure that the trap handler returns the correct exit
code.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The function is only used in do_estimate. Move it to kdumpctl to
prevent confusion.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Pingfan Liu <piliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The function is pretty broken right now. To start with the -o/--omit
option allows a quoted, space separated list of modules. But using 'set'
breaks quotation and thus only considers the first element in the list.
Furthermore dracut uses getopt internally. This means that it is also
possible to pass the list via --omit=.
Fix the function by making use of getopt for parsing the dracut_args.
While at it also add a test cases to cover the functions.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The function is only used in kdumpctl. Thus move it there to keep
kdump-lib small and simple.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The function isn't used anywhere. Thus remove it to keep kdump-lib small
and simple.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The function isn't used anywhere. Thus remove it to keep kdump-lib small
and simple.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
With the NetworkManager rewrite this function in no longer used. This
also allows to remove a lot of dead code in kdump-lib.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
The function isn't used anywhere. Thus remove it to keep
kdump-lib-initramfs small and simple.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Lichen Liu <lichliu@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
If the device/mountpoint for findmnt is omitted findmnt will list all
mounted filesystems. In that case it will always return "true". So
explicitly check if an argument was passed to prevent false-positives.
Signed-off-by: Philipp Rudo <prudo@redhat.com>
Reviewed-by: Coiby Xu <coxu@redhat.com>
Relates: https://bugzilla.redhat.com/show_bug.cgi?id=2151504
Currently, when the network isn't ready, kdump would repeatedly print
the same info,
[ 29.537230] kdump[671]: Bad kdump network destination: 192.123.1.21
[ 30.559418] kdump[679]: Bad kdump network destination: 192.123.1.21
[ 31.580189] kdump[687]: Bad kdump network destination: 192.123.1.21
This is not user-friendly and users may think kdump has got stuck. So
also show much time has waited for the network to be ready,
[ 29.546258] kdump[673]: Waiting for network to be ready (50s / 10min)
...
[ 32.608967] kdump[697]: Waiting for network to be ready (56s / 10min)
Note kdump_get_ip_route no longer prints an error message and it's up to
the caller to determine the log level and print relevant messages. And
kdump_collect_netif_usage aborts when kdump_get_ip_route fails.
Reported-by: Martin Pitt <mpitt@redhat.com>
Signed-off-by: Coiby Xu <coxu@redhat.com>
Reviewed-by: Philipp Rudo <prudo@redhat.com>
