Re: [PATCH 2/7] kprobes: Add kprobe_insn_mutex and cleanup arch_remove_kprobe()

[Masami Hiramatsu <mhiramat@redhat.com>]

Hi Andrew,

Andrew Morton wrote:
> On Tue, 11 Nov 2008 15:56:21 -0500
> Masami Hiramatsu <mhiramat@redhat.com> wrote:
> 
>> Add kprobe_insn_mutex for protecting kprobe_insn_pages hlist, and remove
>> kprobe_mutex from architecture dependent code.
>>
>> This allows us to call arch_remove_kprobe() (and free_insn_slot) while holding
>> kprobe_mutex.
>>
>> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
>> ---
>>  arch/arm/kernel/kprobes.c     |    2 --
>>  arch/ia64/kernel/kprobes.c    |    8 +++++---
>>  arch/powerpc/kernel/kprobes.c |    7 ++++---
>>  arch/s390/kernel/kprobes.c    |    7 ++++---
>>  arch/x86/kernel/kprobes.c     |    7 ++++---
>>  include/linux/kprobes.h       |    1 -
>>  kernel/kprobes.c              |   25 +++++++++++++++++++++----
> 
> For some reasons sparc64 never had this mutex.  You've checked that
> sparc64 will be OK after this change?

I think it's OK for those archs.

Actually, this mutex is protecting insn_slot which stores copied
instructions into executable memory. Some architectures which
doesn't need that special memory have its own buffer in the
arch_specific_insn and they don't use insn_slot. On those arch,
kprobe_insn_mutex is not used (not defined).

Thank you,

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com