If MCE decoding support does not exist for a particular family/model,
and if one tries to inject errors using mce_amd_inj module, it leads
to kernel OOPS. Especially if we inject errors to MC0, MC1, MC2 banks.
Sample:
[ 60.567808] [Hardware Error]: MC0 Error:
[ 60.567826] BUG: unable to handle kernel NULL pointer dereference at
(null)
[ 60.567840] IP: [<ffffffffa0019f36>] amd_decode_mce+0x526/0x900
[edac_mce_amd]
[ 60.567855] PGD ba665067 PUD 37168067 PMD 0
[ 60.567865] Oops: 0000 [#1] SMP
[ 60.567872] Modules linked in: mce_amd_inj amd64_edac_mod edac_core
edac_mce_amd r8169
[ 60.567889] CPU: 2 PID: 2011 Comm: sh Not tainted 3.14.0-rc3.spinoff_ML+ #7
[ 60.567898] Hardware name: AMD Lamar/Lamar, BIOS WLA3904N_Weekly_13_09_0
09/04/2013
[ 60.567907] task: ffff88040a58e040 ti: ffff8800bb206000 task.ti:
ffff8800bb206000
[ 60.567916] RIP: 0010:[<ffffffffa0019f36>] [<ffffffffa0019f36>]
amd_decode_mce+0x526/0x900 [edac_mce_amd]
[ 60.567930] RSP: 0018:ffff8800bb207dc8 EFLAGS: 00010206
[ 60.567937] RAX: 0000000000000000 RBX: ffffffffa0014300 RCX: 00000000000010a5
[ 60.567945] RDX: 0000000000002825 RSI: 0000000000000001 RDI: 0000000000000f0f
[ 60.567953] RBP: ffff8800bb207e48 R08: 0000000000000000 R09: 0000000000000370
[ 60.567961] R10: ffffffff81a6ace0 R11: f000000000000000 R12: 0000000000012980
[ 60.567968] R13: a000000000010f0f R14: 0000000000000001 R15: ffff88041fc00000
[ 60.567978] FS: 00007f4709286700(0000) GS:ffff88041fd00000(0000)
knlGS:0000000000000000
[ 60.567988] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 60.567995] CR2: 0000000000000000 CR3: 00000000bb1b3000 CR4: 00000000000407e0
[ 60.568003] Stack:
[ 60.568007] ffffea0000daa000 0000000000000000 ffff88040a592020
0000000000c504a8
[ 60.568021] ffff880409e7b6d0 ffff8800bb207e60 ffff88040d99aba0
ffff8800bb207e38
[ 60.568033] ffffffff813ce4fc 0000000a00000006 0000000000c504a8
0000000000000002
[ 60.568045] Call Trace:
[ 60.568059] [<ffffffff813ce4fc>] ? _kstrtoull+0x2c/0x90
[ 60.568069] [<ffffffffa0012056>] edac_inject_bank_store+0x56/0x90
[mce_amd_inj]
[ 60.568083] [<ffffffff81272b10>] ? kernfs_fop_write+0x50/0x150
[ 60.568094] [<ffffffff813bc56f>] kobj_attr_store+0xf/0x20
[ 60.568104] [<ffffffff8126ee25>] sysfs_kf_write+0x45/0x60
[ 60.568114] [<ffffffff81272b9e>] kernfs_fop_write+0xde/0x150
[ 60.568125] [<ffffffff811fb422>] vfs_write+0xc2/0x1d0
[ 60.568134] [<ffffffff811fb8f2>] SyS_write+0x52/0xa0
[ 60.568144] [<ffffffff8184361e>] ? do_page_fault+0xe/0x10
[ 60.568154] [<ffffffff81848512>] system_call_fastpath+0x16/0x1b
[ 60.568162] Code: c7 17 ba 01 a0 31 c0 4a 8b 34 ed e0 b3 01 a0 e8 12 64 81
e1 4c 8b 2b e9 3f fb ff ff 48 8b 05 7a 33 00 00 41 0f b6 f6 41 0f b7 fd <ff> 10
84 c0 0f 85 fc fd ff ff 48 c7 c7 28 c3 01 a0 31 c0 e8 e3
[ 60.568228] RIP [<ffffffffa0019f36>] amd_decode_mce+0x526/0x900
[edac_mce_amd]
[ 60.568240] RSP <ffff8800bb207dc8>
[ 60.568245] CR2: 0000000000000000
[ 60.568252] ---[ end trace 6ba951fb82ecbc10 ]---
In this patch, we fix the bug by checking if fam_ops struct has been
alloc-ed before we proceed with fam/model specific decoding.
Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrish...@amd.com>
---
drivers/edac/mce_amd.c | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 30f7309..9b03daa 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -281,6 +281,12 @@ static void decode_mc0_mce(struct mce *m)
pr_emerg(HW_ERR "MC0 Error: ");
+ if (!fam_ops) {
+ pr_err("fam_ops structure not alloc-ed."
+ " Cannot provide detailed family/model"
+ " specific error decoding.\n");
+ return;
+ }
/* TLB error signatures are the same across families */
if (TLB_ERROR(ec)) {
if (TT(ec) == TT_DATA) {
@@ -391,6 +397,12 @@ static void decode_mc1_mce(struct mce *m)
pr_emerg(HW_ERR "MC1 Error: ");
+ if (!fam_ops) {
+ pr_err("fam_ops structure not alloc-ed."
+ " Cannot provide detailed family/model"
+ " specific error decoding.\n");
+ return;
+ }
if (TLB_ERROR(ec))
pr_cont("%s TLB %s.\n", LL_MSG(ec),
(xec ? "multimatch" : "parity error"));
@@ -522,6 +534,12 @@ static void decode_mc2_mce(struct mce *m)
pr_emerg(HW_ERR "MC2 Error: ");
+ if (!fam_ops) {
+ pr_err("fam_ops structure not alloc-ed."
+ " Cannot provide detailed family/model"
+ " specific error decoding.\n");
+ return;
+ }
if (!fam_ops->mc2_mce(ec, xec))
pr_cont(HW_ERR "Corrupted MC2 MCE info?\n");
}
--
1.7.10.4
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
