Messages in this thread | | | Date | Mon, 17 Feb 2014 13:26:20 -0600 | From | Aravind Gopalakrishnan <> | Subject | Re: [PATCH] EDAC, MCE, AMD: Fix code to prevent NULL dereference |
| |
On 2/17/2014 12:27 PM, Borislav Petkov wrote: > On Mon, Feb 17, 2014 at 11:49:51AM -0600, Aravind Gopalakrishnan wrote: >> 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. > Well, we shouldn't even be loading the module on unsupported hw, i.e., > something like that: > > > if (c->x86_vendor != X86_VENDOR_AMD) > - return 0; > + return -ENODEV; > > if (c->x86 < 0xf || c->x86 > 0x16) > - return 0; > + return -ENODEV; > Ah. you are right. (I had tested against different family/model checks that happen to throw NULL dereference) You can ignore the patch.
Apologies for the trouble.
-Aravind.
| |