Messages in this thread |  | | | Date | Fri, 20 Jun 2008 08:29:58 +0200 | | From | Ingo Molnar <> | | Subject | Re: [PATCH] x86: make 64bit identify_cpu use cpu_dev v2 |
| |
* Yinghai Lu <yhlu.kernel@gmail.com> wrote:
> v2: fix early_panic on some conf > reason : struct cpu_vendor_dev size is 16, need to make table to be 16 > byte alignment > also print out the cpu supported...
applied to tip/x86/cpu, thanks Yinghai.
> --- linux-2.6.orig/arch/x86/kernel/vmlinux_64.lds.S > +++ linux-2.6/arch/x86/kernel/vmlinux_64.lds.S > @@ -168,6 +168,7 @@ SECTIONS > *(.con_initcall.init) > } > __con_initcall_end = .; > + . = ALIGN(16); > __x86cpuvendor_start = .; > .x86cpuvendor.init : AT(ADDR(.x86cpuvendor.init) - LOAD_OFFSET) { > *(.x86cpuvendor.init)
ah, nice one!
we broke that via v2.6.25-37-g03ae576 "x86: use ELF section to list CPU vendor specific code" - should we backport your fix to mainline right now? I'm wondering why it only triggered now - did alignment change in this section due to your changes and thus this dormant bug became triggerable?
Ingo
|  |