RE: [tip:x86/efi2] efivars: efivar_entry API

[Seiji Aguchi]

Matt,

Thanks.
With your patch, It works in case each entry is erased one by one as below.
# rm dmesg-efi-1
#rm dmesg-efi-2 

But, it still panics in case multiple entries are erased at the same time as below.
#rm dmsg-efi-*

SELinux: initialized (dev pstore, type pstore), not configured for labeling
BUG: unable to handle kernel NULL pointer dereference at           (null)
IP: [<ffffffff8142ea0f>] __efivar_entry_iter+0xcf/0x120
PGD 19483f067 PUD 195426067 PMD 0 
Oops: 0000 [#1] SMP 
Modules linked in: ebtable_nat ebtables xt_CHECKSUM iptable_mangle bridge autofs4 sunrpc 8021q garp stp llc cpufreq_ondemand ipt_REJECT nf_conntrack_ipv4 nf_defrag_ipv4 iptable_filter ip_tables ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables ipv6 vfat fat vhost_net macvtap macvlan tun uinput thinkpad_acpi iTCO_wdt iTCO_vendor_support wmi sg acpi_cpufreq freq_table mperf arc4 coretemp kvm_intel kvm iwldvm mac80211 crc32c_intel ghash_clmulni_intel aesni_intel ablk_helper cryptd lrw aes_x86_64 xts gf128mul microcode pcspkr i2c_i801 lpc_ich mfd_core iwlwifi cfg80211 rfkill snd_hda_codec_hdmi snd_hda_codec_conexant snd_hda_intel snd_hda_codec snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore snd_page_alloc e1000e ptp pps_core ext4(F) mbcache(F) jbd2(F) sd_mod(F) crc_t10dif(F) sdhci_pci(F) sdhci(F) mmc_core(F) ahci(F) libahci(F) i915(F) drm_kms_helper(F) drm(F) i2c_algo_bit(F) i2c_core(F) video(F) dm_mirror(F) dm_region_!
 hash(F) dm
_log(F) dm_mod(F)
CPU 3 
Pid: 13472, comm: rm Tainted: GF            3.9.0-rc8-tip+ #6 LENOVO 4291EV7/4291EV7
RIP: 0010:[<ffffffff8142ea0f>]  [<ffffffff8142ea0f>] __efivar_entry_iter+0xcf/0x120
RSP: 0018:ffff880194395ca8  EFLAGS: 00010046
RAX: 0000000000000000 RBX: ffffffff81ab8de0 RCX: 000000000000000f
RDX: 0000000000000000 RSI: ffff880194395c59 RDI: ffff880194395c49
RBP: ffff880194395ce8 R08: 000000000000fff2 R09: 000000000000000a
R10: 0000000000000000 R11: 000000000000fff5 R12: ffffffff81430f10
R13: ffff880194395d88 R14: fffffffffffff7d8 R15: ffff880194395db0
FS:  00007f6e8afb4700(0000) GS:ffff88019e2c0000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000000 CR3: 0000000194915000 CR4: 00000000000407e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process rm (pid: 13472, threadinfo ffff880194394000, task ffff88019107f4e0)
Stack:
 ffff880194395cb8 ffff880195bc1000 ffff880194395cc8 000000005177ef75
 0000000000000000 000000000000000a 0000000000000000 0000000000000001
 ffff880194395e28 ffffffff81430ebf ffff88019107f4e0 ffff880194395db8
Call Trace:
 [<ffffffff81430ebf>] efi_pstore_erase+0xef/0x140
 [<ffffffff81003138>] ? math_error+0x288/0x2d0
 [<ffffffff811ea491>] pstore_unlink+0x41/0x60
 [<ffffffff811741ff>] vfs_unlink+0x9f/0x110
 [<ffffffff8117813b>] do_unlinkat+0x18b/0x280
 [<ffffffff8116d7e6>] ? sys_newfstatat+0x36/0x50
 [<ffffffff81178472>] sys_unlinkat+0x22/0x40
 [<ffffffff81543282>] system_call_fastpath+0x16/0x1b
Code: 8d 82 d8 f7 ff ff 48 89 45 c8 4c 8b b0 28 08 00 00 31 c0 48 39 d3 74 38 49 81 ee 28 08 00 00 eb 21 0f 1f 00 49 8d 96 28 08 00 00 <49> 8b 8e 28 08 00 00 48 39 d3 74 35 4c 89 75 c8 4c 8d b1 d8 f7 
RIP  [<ffffffff8142ea0f>] __efivar_entry_iter+0xcf/0x120
 RSP <ffff880194395ca8>
CR2: 0000000000000000
---[ end trace 1d19d659e0c71627 ]---

> -----Original Message-----
> From: Matt Fleming [mailto:matt.fleming@xxxxxxxxx]
> Sent: Friday, April 26, 2013 5:56 AM
> To: Seiji Aguchi
> Cc: linux-kernel@xxxxxxxxxxxxxxx; mingo@xxxxxxxxxx; hpa@xxxxxxxxx; mjg59@xxxxxxxxxxxxx; tony.luck@xxxxxxxxx; jk@xxxxxxxxxx;
> teg@xxxxxxx; tglx@xxxxxxxxxxxxx; mikew@xxxxxxxxxx; linux-tip-commits@xxxxxxxxxxxxxxx
> Subject: Re: [tip:x86/efi2] efivars: efivar_entry API
> 
> On 24/04/13 00:55, Seiji Aguchi wrote:
> > Hi,
> >
> > I tested a current tip tree to check if the new API works.
> > But pstore_erase() doesn't work...
> > I'm checking the source code right now.
> >
> > Seiji
> 
> [...]
> 
> > Call Trace:
> >  [<ffffffff8143001f>] efi_pstore_erase+0xdf/0x130
> >  [<ffffffff81200038>] ? cap_socket_create+0x8/0x10
> >  [<ffffffff811ea491>] pstore_unlink+0x41/0x60
> >  [<ffffffff811741ff>] vfs_unlink+0x9f/0x110
> >  [<ffffffff8117813b>] do_unlinkat+0x18b/0x280
> >  [<ffffffff81178472>] sys_unlinkat+0x22/0x40
> >  [<ffffffff81542402>] system_call_fastpath+0x16/0x1b
> 
> Does this patch fix things?
> 
> ---
> 
> diff --git a/drivers/firmware/efi/efi-pstore.c b/drivers/firmware/efi/efi-pstore.c
> index 47ae712..b820593 100644
> --- a/drivers/firmware/efi/efi-pstore.c
> +++ b/drivers/firmware/efi/efi-pstore.c
> @@ -173,7 +173,7 @@ static int efi_pstore_erase(enum pstore_type_id type, u64 id, int count,
>  			    struct timespec time, struct pstore_info *psi)
>  {
>  	struct pstore_erase_data edata;
> -	struct efivar_entry *entry;
> +	struct efivar_entry *entry = NULL;
>  	char name[DUMP_NAME_LEN];
>  	efi_char16_t efi_name[DUMP_NAME_LEN];
>  	int found, i;
N‹§²æìr¸›yúèšØb²X¬¶ÇvØ^–)Þ{.nÇ+‰·¥Š{±‘êçzX§¶›¡Ü}©ž²ÆzÚ&j:+v‰¨¾«‘êçzZ+€Ê+zf£¢·hšˆ§~†­†Ûiÿûàz¹®w¥¢¸?™¨è­Ú&¢)ßf”ù^jÇy§m…á@A«a¶Úÿ0¶ìh®å’i