Messages in this thread | | | Date | Tue, 28 Jan 2014 15:43:57 -0500 | From | Boris Ostrovsky <> | Subject | Re: AMD microcode loading broken on 32 bit |
| |
On 01/28/2014 11:24 AM, Borislav Petkov wrote: > On Thu, Jan 23, 2014 at 02:41:59PM -0500, Boris Ostrovsky wrote: >> 32-bit only. > Ok, I think I know what happens. Can you try this one (I missed doing > this on 32-bit and 64-bit does it, which would explain why it didn't > explode there): > > --- > diff --git a/arch/x86/kernel/cpu/microcode/amd_early.c b/arch/x86/kernel/cpu/microcode/amd_early.c > index 8384c0fa206f..03b759401bc1 100644 > --- a/arch/x86/kernel/cpu/microcode/amd_early.c > +++ b/arch/x86/kernel/cpu/microcode/amd_early.c > @@ -359,7 +359,7 @@ int __init save_microcode_in_initrd_amd(void) > pr_info("microcode: updated early to new patch_level=0x%08x\n", > ucode_new_rev); > > - if (!container) > + if (!container || !ucode_cpio.data) > return -EINVAL; > > eax = cpuid_eax(0x00000001);
It fixes the case when there is no microcode in initrd but when microcode is corrupted (as was the case when we were pointing to Intel binary) we still die. Neither container nor ucode_cpio.data is NULL in this case.
-boris
| |