[PATCH] fix kvm kdump regression

[PATCH] fix kvm kdump regression

[caiqian]

020ec6537aa65c18e9084c568d7b94727f2026fd introduced a regression that an empty vmcore was generated during kdump on a kvm guest.
ioremap: invalid physical address db74000000000000
------------[ cut here ]------------
WARNING: at arch/x86/mm/ioremap.c:83 __ioremap_caller+0x307/0x380()
Hardware name: KVM
Modules linked in:
Pid: 1, comm: swapper Not tainted 2.6.36-rc2-mm1+ #2
Call Trace:
 [<ffffffff8105f48f>] warn_slowpath_common+0x7f/0xc0
 [<ffffffff8103052e>] ? copy_oldmem_page+0x4e/0xc0
 [<ffffffff8105f4ea>] warn_slowpath_null+0x1a/0x20
 [<ffffffff8103f0a7>] __ioremap_caller+0x307/0x380
 [<ffffffff8103f1f7>] ioremap_nocache+0x17/0x20
 [<ffffffff8103052e>] copy_oldmem_page+0x4e/0xc0
 [<ffffffff811af4ca>] read_from_oldmem+0x7a/0xb0
 [<ffffffff81c4f70b>] merge_note_headers_elf64.clone.1+0x6c/0x21f
 [<ffffffff8103056e>] ? copy_oldmem_page+0x8e/0xc0
 [<ffffffff811af4ca>] ? read_from_oldmem+0x7a/0xb0
 [<ffffffff81c4fa5b>] vmcore_init+0x19d/0x396
 [<ffffffff81c4f8be>] ? vmcore_init+0x0/0x396
 [<ffffffff81002053>] do_one_initcall+0x43/0x190
 [<ffffffff81c278ab>] kernel_init+0x2a0/0x330
 [<ffffffff8100be84>] kernel_thread_helper+0x4/0x10
 [<ffffffff81c2760b>] ? kernel_init+0x0/0x330
 [<ffffffff8100be80>] ? kernel_thread_helper+0x0/0x10
---[ end trace 93d72a36b9146f22 ]---
Kdump: vmcore not initialized

This patch fixed it by reverting a chunk of the above commit and 9983b6f0cf8263e51bcf4c8a9dc0c1ef175b3c60 which is based on it.

Signed-off-by: CAI Qian <caiqian@redhat.com>

diff --git a/mm/percpu.c b/mm/percpu.c
index 77e3f5a..0dbb12a 100644
--- a/mm/percpu.c
+++ b/mm/percpu.c
@@ -986,38 +986,10 @@ bool is_kernel_percpu_address(unsigned long addr)
  */
 phys_addr_t per_cpu_ptr_to_phys(void *addr)
 {
-	void __percpu *base = __addr_to_pcpu_ptr(pcpu_base_addr);
-	bool in_first_chunk = false;
-	unsigned long first_start, first_end;
-	unsigned int cpu;
-
-	/*
-	 * The following test on first_start/end isn't strictly
-	 * necessary but will speed up lookups of addresses which
-	 * aren't in the first chunk.
-	 */
-	first_start = pcpu_chunk_addr(pcpu_first_chunk, pcpu_first_unit_cpu, 0);
-	first_end = pcpu_chunk_addr(pcpu_first_chunk, pcpu_last_unit_cpu,
-				    pcpu_unit_pages);
-	if ((unsigned long)addr >= first_start &&
-	    (unsigned long)addr < first_end) {
-		for_each_possible_cpu(cpu) {
-			void *start = per_cpu_ptr(base, cpu);
-
-			if (addr >= start && addr < start + pcpu_unit_size) {
-				in_first_chunk = true;
-				break;
-			}
-		}
-	}
-
-	if (in_first_chunk) {
-		if ((unsigned long)addr < VMALLOC_START ||
-		    (unsigned long)addr >= VMALLOC_END)
-			return __pa(addr);
-		else
-			return page_to_phys(vmalloc_to_page(addr));
-	} else
+	if ((unsigned long)addr < VMALLOC_START ||
+			(unsigned long)addr >= VMALLOC_END)
+		return __pa(addr);
+	else
 		return page_to_phys(pcpu_addr_to_page(addr));
 }

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] fix kvm kdump regression

[Tejun Heo]

On 09/16/2010 06:57 AM, caiqian@redhat.com wrote:
> 020ec6537aa65c18e9084c568d7b94727f2026fd introduced a regression that an empty vmcore was generated during kdump on a kvm guest.
> ioremap: invalid physical address db74000000000000
> ------------[ cut here ]------------
> WARNING: at arch/x86/mm/ioremap.c:83 __ioremap_caller+0x307/0x380()
> Hardware name: KVM
> Modules linked in:
> Pid: 1, comm: swapper Not tainted 2.6.36-rc2-mm1+ #2
> Call Trace:
>  [<ffffffff8105f48f>] warn_slowpath_common+0x7f/0xc0
>  [<ffffffff8103052e>] ? copy_oldmem_page+0x4e/0xc0
>  [<ffffffff8105f4ea>] warn_slowpath_null+0x1a/0x20
>  [<ffffffff8103f0a7>] __ioremap_caller+0x307/0x380
>  [<ffffffff8103f1f7>] ioremap_nocache+0x17/0x20
>  [<ffffffff8103052e>] copy_oldmem_page+0x4e/0xc0
>  [<ffffffff811af4ca>] read_from_oldmem+0x7a/0xb0
>  [<ffffffff81c4f70b>] merge_note_headers_elf64.clone.1+0x6c/0x21f
>  [<ffffffff8103056e>] ? copy_oldmem_page+0x8e/0xc0
>  [<ffffffff811af4ca>] ? read_from_oldmem+0x7a/0xb0
>  [<ffffffff81c4fa5b>] vmcore_init+0x19d/0x396
>  [<ffffffff81c4f8be>] ? vmcore_init+0x0/0x396
>  [<ffffffff81002053>] do_one_initcall+0x43/0x190
>  [<ffffffff81c278ab>] kernel_init+0x2a0/0x330
>  [<ffffffff8100be84>] kernel_thread_helper+0x4/0x10
>  [<ffffffff81c2760b>] ? kernel_init+0x0/0x330
>  [<ffffffff8100be80>] ? kernel_thread_helper+0x0/0x10
> ---[ end trace 93d72a36b9146f22 ]---
> Kdump: vmcore not initialized
> 
> This patch fixed it by reverting a chunk of the above commit and 9983b6f0cf8263e51bcf4c8a9dc0c1ef175b3c60 which is based on it.
> 
> Signed-off-by: CAI Qian <caiqian@redhat.com>

Sorry about the delay but I'm prepping a debug patch.  Let's root
cause and see if it can be fixed without reverting it as there are
further changes which depend on the commit and we'll have to hunt it
down anyway.

Thanks.

-- 
tejun

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec