Re: [tip:x86/urgent] x86/kasan: Fix boot with KASAN=y and PROFILE_ANNOTATED_BRANCHES=y

[Andrey Ryabinin]

On 03/14/2017 05:59 PM, Andrey Ryabinin wrote:
> On 03/14/2017 05:09 PM, Alexander Potapenko wrote:
>> On Tue, Mar 14, 2017 at 12:03 AM, tip-bot for Andrey Ryabinin
>> <tipbot@xxxxxxxxx> wrote:
>>> Commit-ID:  be3606ff739d1c1be36389f8737c577ad87e1f57
>>> Gitweb:     http://git.kernel.org/tip/be3606ff739d1c1be36389f8737c577ad87e1f57
>>> Author:     Andrey Ryabinin <aryabinin@xxxxxxxxxxxxx>
>>> AuthorDate: Mon, 13 Mar 2017 19:33:37 +0300
>>> Committer:  Thomas Gleixner <tglx@xxxxxxxxxxxxx>
>>> CommitDate: Tue, 14 Mar 2017 00:00:55 +0100
>>>
>>> x86/kasan: Fix boot with KASAN=y and PROFILE_ANNOTATED_BRANCHES=y
>>>
>>> The kernel doesn't boot with both PROFILE_ANNOTATED_BRANCHES=y and KASAN=y
>>> options selected. With branch profiling enabled we end up calling
>>> ftrace_likely_update() before kasan_early_init(). ftrace_likely_update() is
>>> built with KASAN instrumentation, so calling it before kasan has been
>>> initialized leads to crash.
>>>
>>> Use DISABLE_BRANCH_PROFILING define to make sure that we don't call
>>> ftrace_likely_update() from early code before kasan_early_init().
>> Won't it suffice to disable instrumentation of ftrace_likely_update() and alike?
> 
> Not just ftrace_likely_update(), but also all functions that ftrace_likely_update() might call.
> That would suffice, but it would be unreasonable way to fix this, given that we can just
>       ftrace_likely_update() call.

 ^^avoid

> Plus we don't have any sane way to disable instrumentation per-function.
>