----- "Mahesh J Salgaonkar" <mahesh(a)linux.vnet.ibm.com> wrote:
Hi All,
There was problem in my earlier posting. I had miss-spelled my id in the
From: field. Hence resending it again.
Please find the version 3 of crash patchset that enables crash utility to read
s390x diskdump (DUMPFILE generated by makedumpfile).
The first patch enhances crash utility to read elf notes section from
DUMPFILE if Kdump header >= 4.
The second patch adds crash support to read s390x diskdump. This patch also
implements new arch dependent interface process_elf_notes() to read CPU
register infromation stored in ELF notes section.
Change v3:
- Re-based to crash-5.0.9
Change v2:
- Re-based to crash-5.0.8
- Modified __diskdump_memory_dump() function to display contents of
all 4 additional fields in the kdump_sub_header.
- Modified __diskdump_memory_dump() to dump the vmcoreinfo ASCII strings.
Please review the changes and send in your comments.
Hello Mahesh,
The patch-set looks good so far, except for one minor nit.
Why is "S390" accepted as legitimate here in read_dump_header()?
@@ -349,6 +352,10 @@ restart:
dd->machine_type = EM_IA_64;
else if (machine_type("PPC64"))
dd->machine_type = EM_PPC64;
+ else if (machine_type("S390"))
+ dd->machine_type = EM_S390;
+ else if (machine_type("S390X"))
+ dd->machine_type = EM_S390;
else {
error(INFO, "%s: unsupported machine type: %s\n",
DISKDUMP_VALID() ? "diskdump" : "compressed
kdump",
In any case, the only thing I can test at this point is that the
vmcoreinfo ASCII data gets dumped -- and thanks for adding that BTW.
When I get access to, and test with, your sample s390x vmlinux/vmcore
pairs that I requested off-list, I'll confirm that the patch is queued
for the next release.
Thanks,
Dave