Hi Atsushi, Happy new year! On 12/26/17 at 08:21am, Atsushi Kumagai wrote: > >On 12/22/17 at 12:54pm, Dave Young wrote: > >> Hi Atsushi, > >> On 12/21/17 at 08:48am, Atsushi Kumagai wrote: > >> > Hello Dave, > >> > > >> > >[dyoung at dhcp-*-* makedumpfile]$ sudo ./makedumpfile -l -d 31 /mnt/vmcore/vmcore /tmp/vmcore.1 > >> > >The kernel version is not supported. > >> > >The makedumpfile operation may be incomplete. > >> > >Checking for memory holes : [100.0 %] | __vtop4_x86_64: Can't get a valid > >pte. > >> > >readmem: Can't convert a virtual address(ffff88007ebb1000) to physical address. > >> > >readmem: type_addr: 0, addr:ffff88007ebb1000, size:32768 > >> > >__exclude_unnecessary_pages: Can't read the buffer of struct page. > >> > >create_2nd_bitmap: Can't exclude unnecessary pages. > >> > > > >> > >makedumpfile Failed. > >> > > > >> > >If you need the vmcore for debugging please let me know, my test is just > >> > >a normal test in kvm guest. > >> > > >> > Thanks for your report. > >> > I can't reproduce that in my environment, could you give me the vmcore ? > >> > It's helpful if you append the vmlinux and the .config. > >> > >> I'm bisecting it, nearly finished, will post the results and share > >> vmcore if needed soon. > > > >vmcore/vmlinux (gzipped) and .config: > >http://people.redhat.com/ruyang/rhbz1528542/ > > Thanks. > > >Bisect result is below: > > I'm not sure why this commit affects makedumpfile yet, but > I'll give priority to release v1.6.3 since the commit will not be included > in the supported kernel(4.14). > > Regards, > Atsushi Kumagai > > >83e3c48729d9ebb7af5a31a504f3fd6aff0348c4 is the first bad commit > >commit 83e3c48729d9ebb7af5a31a504f3fd6aff0348c4 > >Author: Kirill A. Shutemov <kirill.shutemov at linux.intel.com> > >Date: Fri Sep 29 17:08:16 2017 +0300 > > > > mm/sparsemem: Allocate mem_section at runtime for CONFIG_SPARSEMEM_EXTREME=y > > > > Size of the mem_section[] array depends on the size of the physical address space. > > > > In preparation for boot-time switching between paging modes on x86-64 > > we need to make the allocation of mem_section[] dynamic, because otherwise > > we waste a lot of RAM: with CONFIG_NODE_SHIFT=10, mem_section[] size is 32kB > > for 4-level paging and 2MB for 5-level paging mode. > > > > The patch allocates the array on the first call to sparse_memory_present_with_active_regions(). > > > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov at linux.intel.com> > > Cc: Andrew Morton <akpm at linux-foundation.org> > > Cc: Andy Lutomirski <luto at amacapital.net> > > Cc: Borislav Petkov <bp at suse.de> > > Cc: Cyrill Gorcunov <gorcunov at openvz.org> > > Cc: Linus Torvalds <torvalds at linux-foundation.org> > > Cc: Peter Zijlstra <peterz at infradead.org> > > Cc: Thomas Gleixner <tglx at linutronix.de> > > Cc: linux-mm at kvack.org > > Link: http://lkml.kernel.org/r/20170929140821.37654-2-kirill.shutemov at linux.intel.com > > Signed-off-by: Ingo Molnar <mingo at kernel.org> > > > >:040000 040000 68b7ff3eedc2c9ff56e31108f7e982eacbb233fc 0014ee63bebe14efb0e36e0028e2cbe718fd6c30 M include > >:040000 040000 78adbc296527c802400b1f68e0fbd716920726fa a4eea8117cb318527c0d5a6281d68f312f644831 M mm The root cause is this commit makes mem_section as a pointer instead of the static array. VMCOREINFO_SYMBOL() expand it as &mem_section, this is not correct in the test case any more. This hack code works for me: diff --git a/kernel/crash_core.c b/kernel/crash_core.c index b3663896278e..f5fe6068ae39 100644 --- a/kernel/crash_core.c +++ b/kernel/crash_core.c @@ -376,6 +376,8 @@ phys_addr_t __weak paddr_vmcoreinfo_note(void) { return __pa(vmcoreinfo_note); } +#define VMCOREINFO_SYMBOL_HACK(name) \ + vmcoreinfo_append_str("SYMBOL(%s)=%lx\n", #name, (unsigned long)name) static int __init crash_save_vmcoreinfo_init(void) { @@ -410,10 +412,11 @@ static int __init crash_save_vmcoreinfo_init(void) VMCOREINFO_SYMBOL(contig_page_data); #endif #ifdef CONFIG_SPARSEMEM - VMCOREINFO_SYMBOL(mem_section); + VMCOREINFO_SYMBOL_HACK(mem_section); VMCOREINFO_LENGTH(mem_section, NR_SECTION_ROOTS); VMCOREINFO_STRUCT_SIZE(mem_section); VMCOREINFO_OFFSET(mem_section, section_mem_map); + #endif VMCOREINFO_STRUCT_SIZE(page); VMCOREINFO_STRUCT_SIZE(pglist_data); But probably we need this instead, but I can not test it because I do not know how to fix makedumpfile to use a _NUMBER instead of a SYMBOL. Thus bring up the issue, seeking for thoughts and discussion. diff --git a/kernel/crash_core.c b/kernel/crash_core.c index b3663896278e..dfa2238e2c28 100644 --- a/kernel/crash_core.c +++ b/kernel/crash_core.c @@ -410,10 +410,17 @@ static int __init crash_save_vmcoreinfo_init(void) VMCOREINFO_SYMBOL(contig_page_data); #endif #ifdef CONFIG_SPARSEMEM +#ifdef CONFIG_SPARSEMEM_EXTREME + VMCOREINFO_NUMBER(mem_section); +#else VMCOREINFO_SYMBOL(mem_section); +#endif VMCOREINFO_LENGTH(mem_section, NR_SECTION_ROOTS); VMCOREINFO_STRUCT_SIZE(mem_section); VMCOREINFO_OFFSET(mem_section, section_mem_map); [snip] Thanks Dave
