Re: [3.8.0-rc1] BUG in __split_huge_page_pmd() at mm/huge_memory.c:2743( was Re: bug report for 3.8.0-rc1+)

[Hillf Danton]

Hello Peter

[with Andrea and Mel cced]

On Fri, Feb 22, 2013 at 8:51 PM, Peter Hurley <peter@xxxxxxxxxxxxxxxxxx> wrote:
> Hi Kirill,
>
> I thought you might be interested in this.
>
> HEAD is now at a49f0d1... Linux 3.8-rc1
> peter@thor:~/src/kernels/mainline$ sed = mm/huge_memory.c | sed 'N;s/\n/ /' | sed -n '2730,2744p'
> 2730            spin_unlock(&mm->page_table_lock);
> 2731            mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
> 2732            return;
> 2733    }
> 2734    page = pmd_page(*pmd);
> 2735    VM_BUG_ON(!page_count(page));
> 2736    get_page(page);
> 2737    spin_unlock(&mm->page_table_lock);
> 2738    mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
> 2739
> 2740    split_huge_page(page);
> 2741
> 2742    put_page(page);
> 2743    BUG_ON(pmd_trans_huge(*pmd));
> 2744 }
>
Lets check if anon vma still available(see Mel's comment in
split_huge_page()). Diff is based on -next.

Can we remove the trap?

Hillf
---
--- a/mm/huge_memory.c	Sat Feb 23 16:19:40 2013
+++ b/mm/huge_memory.c	Sat Feb 23 16:24:18 2013
@@ -2675,6 +2675,7 @@ static void __split_huge_zero_page_pmd(s
 void __split_huge_page_pmd(struct vm_area_struct *vma, unsigned long address,
 		pmd_t *pmd)
 {
+	int unsplitted;
 	struct page *page;
 	struct mm_struct *mm = vma->vm_mm;
 	unsigned long haddr = address & HPAGE_PMD_MASK;
@@ -2704,10 +2705,11 @@ void __split_huge_page_pmd(struct vm_are
 	spin_unlock(&mm->page_table_lock);
 	mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);

-	split_huge_page(page);
+	unsplitted = split_huge_page(page);

 	put_page(page);
-	BUG_ON(pmd_trans_huge(*pmd));
+	if (!unsplitted)
+		BUG_ON(pmd_trans_huge(*pmd));
 }

 void split_huge_page_pmd_mm(struct mm_struct *mm, unsigned long address,
--

> Regards,
> Peter Hurley
>
>
> On Fri, 2013-02-22 at 12:01 +0200, NKoc wrote:
>> Feb 22 10:44:16 nkoc kernel: ------------[ cut here ]------------
>> Feb 22 10:44:16 nkoc kernel: kernel BUG at mm/huge_memory.c:2743!
>> Feb 22 10:44:16 nkoc kernel: invalid opcode: 0000 [#1] SMP
>> Feb 22 10:44:16 nkoc kernel: Modules linked in: snd_seq_dummy
>> snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device snd_pcm_oss
>> snd_mixer_oss ipv6 ppdev lp parport_pc parport fuse snd_hda_codec_hdmi
>> snd_hda_codec_via snd_hda_intel snd_hda_codec i915 intel_agp snd_hwdep
>> snd_pcm rtl8723ae rtlwifi joydev zd1211rw mac80211 intel_gtt
>> drm_kms_helper hid_generic usbhid hid snd_page_alloc snd_timer cfg80211
>> r8169 btusb drm snd bluetooth acpi_cpufreq mperf agpgart freq_table
>> processor psmouse i2c_i801 thermal rfkill serio_raw i2c_algo_bit video
>> mii i2c_core lpc_ich mei evdev soundcore thermal_sys coretemp mfd_core
>> hwmon wmi button battery ac crc32c_intel microcode loop
>> Feb 22 10:44:16 nkoc kernel: Pid: 2616, comm: thunderbird Tainted:
>> P             3.8.0-rc1+ #1 EXPER KARIZMA/W240EU/W250EUQ/W270EUQ
>> Feb 22 10:44:16 nkoc kernel: EIP: 0060:[<c10e54cc>] EFLAGS: 00210282 CPU: 1
>> Feb 22 10:44:16 nkoc kernel: EIP is at __split_huge_page_pmd+0x1fc/0x220
>> Feb 22 10:44:16 nkoc kernel: EAX: f6934000 EBX: f6934000 ECX: 00000002 EDX: ef42eaf8
>> Feb 22 10:44:16 nkoc kernel: ESI: 00000000 EDI: afbf8000 EBP: 00000000 ESP: f1d75e68
>> Feb 22 10:44:16 nkoc kernel:  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
>> Feb 22 10:44:16 nkoc kernel: CR0: 80050033 CR2: a23fb00c CR3: 2f42e000 CR4: 000407d0
>> Feb 22 10:44:16 nkoc kernel: DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
>> Feb 22 10:44:16 nkoc kernel: DR6: ffff0ff0 DR7: 00000400
>> Feb 22 10:44:16 nkoc kernel: Process thunderbird (pid: 2616, ti=f1d74000 task=f4edc000 task.ti=f1d74000)
>> Feb 22 10:44:16 nkoc kernel: Stack:
>> Feb 22 10:44:16 nkoc kernel:  afc00000 c10b5531 f494d6c0 af800000 ef42eaf8 c10b8568 afbf9000 afbf9000
>> Feb 22 10:44:16 nkoc kernel:  afbf8000 afbf8000 c10ca112 f6630f54 afbf9000 f1d75f14 00000001 00000001
>> Feb 22 10:44:16 nkoc kernel:  f1d45420 ef42eaf8 f4edc000 f1d75ef0 ef42eaf8 afbf9000 afbf8fff a9a63000
>> Feb 22 10:44:16 nkoc kernel: Call Trace:
>> Feb 22 10:44:16 nkoc kernel:  [<c10b5531>] ? free_hot_cold_page_list+0x21/0x40
>> Feb 22 10:44:16 nkoc kernel:  [<c10b8568>] ? release_pages+0x148/0x180
>> Feb 22 10:44:16 nkoc kernel:  [<c10ca112>] ? unmap_single_vma+0xf2/0x560
>> Feb 22 10:44:16 nkoc kernel:  [<c10caf00>] ? zap_page_range+0x80/0xd0
>> Feb 22 10:44:16 nkoc kernel:  [<c107427f>] ? clockevents_program_event+0x9f/0x150
>> Feb 22 10:44:16 nkoc kernel:  [<c10c8310>] ? sys_madvise+0x1f0/0x570
>> Feb 22 10:44:16 nkoc kernel:  [<c10503d0>] ? hrtimer_interrupt+0x170/0x290
>> Feb 22 10:44:16 nkoc kernel:  [<c12c9935>] ? __percpu_counter_add+0x65/0xb0
>> Feb 22 10:44:16 nkoc kernel:  [<c17f142f>] ? syscall_call+0x7/0xb
>> Feb 22 10:44:16 nkoc kernel:  [<c17f0000>] ? __schedule+0x240/0x720
>> Feb 22 10:44:16 nkoc kernel: Code: 4b 69 ff ff 89 d8 e8 64 32 fd ff 84
>> c0 0f 85 c2 fe ff ff e9 b9 fe ff ff 8b 0c 24 8b 54 24 0c e8 bb 68 ff ff
>> e9 50 fe ff ff 0f 0b <0f> 0b 8b 0c 24 8b 54 24 0c 8b 44 24 08 e8 12 69
>> ff ff e9 a9 fe
>> Feb 22 10:44:16 nkoc kernel: EIP: [<c10e54cc>]
>> __split_huge_page_pmd+0x1fc/0x220 SS:ESP 0068:f1d75e68
>> Feb 22 10:44:16 nkoc kernel: ---[ end trace e8618ddbef619686 ]---
>
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/