kexec-tools/kexec-tools-2.0.14-makedumpfile-x86_64-check-physical-address-in-PT_LOA.patch

48 lines
1.8 KiB
Diff
Raw Normal View History

Fix `makedumpfile --mem-usage /proc/kcore` Patches have been taken from kexec-tools and makedumpfile to fix issue with `makedumpfile --mem-usage /proc/kcore`. Two of the patches is from kexec-tools and rest are from makedumpfile. All the patches have been acked upstream and applies without conflict. Kexec-tools patches: (kexec-tools-2.0.14-x86-x86_64-Fix-format-warning-with-die.patch), which fixes koji build issue. kexec-tools-2.0.14-build_mem_phdrs-check-if-p_paddr-is-invalid.patch fixes the regresssion caused by kernel /proc/kcore fix to use -1 as default value of p_paddr for pt_loads. Without his patch kexec -p will fail with latest kernel. Other makedumpfile patches are backported to support --mem-usage while kernel kaslr being enabled. Details please see the patch log of the individual patches. All the patches are backport of upstream commits. Patches has been tested with kernel 4.11.0-0.rc1.git0.1.fc26.x86_64. # makedumpfile --mem-usage /proc/kcore -f The kernel version is not supported. The makedumpfile operation may be incomplete. TYPE PAGES EXCLUDABLE DESCRIPTION ---------------------------------------------------------------------- ZERO 1960 yes Pages filled with zero NON_PRI_CACHE 22850 yes Cache pages without private flag PRI_CACHE 1517 yes Cache pages with private flag USER 32522 yes User process pages FREE 1898981 yes Free pages KERN_DATA 78721 no Dumpable kernel data page size: 4096 Total pages on system: 2036551 Total size on system: 8341712896 Byte We won't need to pass -f once fedora kernel is rebased with v4.12. Signed-off-by: Pratyush Anand <panand@redhat.com> Acked-by: Dave Young <dyoung@redhat.com>
2017-03-14 15:28:30 +00:00
From f1363023b909df886eca5efcb64b78be9b8e6086 Mon Sep 17 00:00:00 2001
Message-Id: <f1363023b909df886eca5efcb64b78be9b8e6086.1489471500.git.panand@redhat.com>
In-Reply-To: <4b0bed3523a5f6c2c428d9dab3d27d4572207d52.1489471500.git.panand@redhat.com>
References: <4b0bed3523a5f6c2c428d9dab3d27d4572207d52.1489471500.git.panand@redhat.com>
From: Pratyush Anand <panand@redhat.com>
Date: Thu, 2 Mar 2017 17:37:13 +0900
Subject: [PATCH 3/7] [PATCH v3 3/7] x86_64: check physical address in PT_LOAD
for none direct mapped regions
A kcore PT_LOAD can have a section from vmalloc region. However,
physical address in that header would be invalid (-1) after kernel
commit 464920104bf7 (/proc/kcore: update physical address for kcore ram
and text). Therefore, check for valid physical address while calculating
page_offset or phys_offset.
Signed-off-by: Pratyush Anand <panand@redhat.com>
---
arch/x86_64.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/makedumpfile-1.6.1/arch/x86_64.c b/makedumpfile-1.6.1/arch/x86_64.c
index 893cd516fc8b..e978a36f8878 100644
--- a/makedumpfile-1.6.1/arch/x86_64.c
+++ b/makedumpfile-1.6.1/arch/x86_64.c
@@ -41,7 +41,8 @@ get_page_offset_x86_64(void)
unsigned long long virt_start;
for (i = 0; get_pt_load(i, &phys_start, NULL, &virt_start, NULL); i++) {
- if (virt_start < __START_KERNEL_map) {
+ if (virt_start < __START_KERNEL_map
+ && phys_start != NOT_PADDR) {
info->page_offset = virt_start - phys_start;
return TRUE;
}
@@ -76,7 +77,8 @@ get_phys_base_x86_64(void)
}
for (i = 0; get_pt_load(i, &phys_start, NULL, &virt_start, NULL); i++) {
- if (virt_start >= __START_KERNEL_map) {
+ if (virt_start >= __START_KERNEL_map
+ && phys_start != NOT_PADDR) {
info->phys_base = phys_start -
(virt_start & ~(__START_KERNEL_map));
--
2.9.3