kexec-tools/kexec-tools-vmcoreinfo-in-xen.patch
Petr Tesařík b9c0cadcb3 Accepting request 593774 from home:ptesarik:branches:Kernel:kdump
- kexec-tools-vmcoreinfo-in-xen.patch: Revert "kexec-tools: Read
  always one vmcoreinfo file" (bsc#1085626, bsc#951740).
- kexec-tools-fix-kexec-p-segfault.patch: Fix a segmentation fault
  when trying to run "kexec -p" (bsc#1080916).

OBS-URL: https://build.opensuse.org/request/show/593774
OBS-URL: https://build.opensuse.org/package/show/Kernel:kdump/kexec-tools?expand=0&rev=97
2018-04-05 11:51:19 +00:00

80 lines
2.5 KiB
Diff

From: Petr Tesarik <petr@tesarici.cz>
Date: Thu, 5 Apr 2018 10:57:05 +0200
Subject: Revert "kexec-tools: Read always one vmcoreinfo file"
References: bsc#1085626, bsc#951740
Upstream: not yet, upstream wants to stay broken
This reverts commit 455d79f57e9367e5c59093fd74798905bd5762fc.
The explanation seems bogus. The file under /sys/kernel
refers to the Linux kernel VMCOREINFO note, while the
file under /sys/hypervisor refers to the Xen hypervisor
VMCOREINFO_XEN note. The former note contains information
about the Linux kernel. The latter contains information
about the Xen hypervisor. Both are needed to allow page
filtering in makedumpfile.
Signed-off-by: Petr Tesarik <ptesarik@suse.cz>
---
kexec/crashdump-elf.c | 33 ++++++++++++++++++++++++++-------
1 file changed, 26 insertions(+), 7 deletions(-)
--- a/kexec/crashdump-elf.c
+++ b/kexec/crashdump-elf.c
@@ -40,6 +40,8 @@ int FUNC(struct kexec_info *info,
uint64_t notes_addr, notes_len;
uint64_t vmcoreinfo_addr, vmcoreinfo_len;
int has_vmcoreinfo = 0;
+ uint64_t vmcoreinfo_addr_xen, vmcoreinfo_len_xen;
+ int has_vmcoreinfo_xen = 0;
int (*get_note_info)(int cpu, uint64_t *addr, uint64_t *len);
long int count_cpu;
@@ -52,14 +54,16 @@ int FUNC(struct kexec_info *info,
return -1;
}
- if (xen_present()) {
- if (!get_xen_vmcoreinfo(&vmcoreinfo_addr, &vmcoreinfo_len))
- has_vmcoreinfo = 1;
- } else
- if (!get_kernel_vmcoreinfo(&vmcoreinfo_addr, &vmcoreinfo_len))
- has_vmcoreinfo = 1;
+ if (get_kernel_vmcoreinfo(&vmcoreinfo_addr, &vmcoreinfo_len) == 0) {
+ has_vmcoreinfo = 1;
+ }
+
+ if (xen_present() &&
+ get_xen_vmcoreinfo(&vmcoreinfo_addr_xen, &vmcoreinfo_len_xen) == 0) {
+ has_vmcoreinfo_xen = 1;
+ }
- sz = sizeof(EHDR) + (nr_cpus + has_vmcoreinfo) * sizeof(PHDR) +
+ sz = sizeof(EHDR) + (nr_cpus + has_vmcoreinfo + has_vmcoreinfo_xen) * sizeof(PHDR) +
ranges * sizeof(PHDR);
/*
@@ -178,6 +182,21 @@ int FUNC(struct kexec_info *info,
dbgprintf_phdr("vmcoreinfo header", phdr);
}
+ if (has_vmcoreinfo_xen) {
+ phdr = (PHDR *) bufp;
+ bufp += sizeof(PHDR);
+ phdr->p_type = PT_NOTE;
+ phdr->p_flags = 0;
+ phdr->p_offset = phdr->p_paddr = vmcoreinfo_addr_xen;
+ phdr->p_vaddr = 0;
+ phdr->p_filesz = phdr->p_memsz = vmcoreinfo_len_xen;
+ /* Do we need any alignment of segments? */
+ phdr->p_align = 0;
+
+ (elf->e_phnum)++;
+ dbgprintf_phdr("vmcoreinfo_xen header", phdr);
+ }
+
/* Setup an PT_LOAD type program header for the region where
* Kernel is mapped if elf_info->kern_size is non-zero.
*/