linux-mips
[Top] [All Lists]

Re: missing data cache flush for signal trampoline on fork

To: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Subject: Re: missing data cache flush for signal trampoline on fork
From: Ralf Baechle <ralf@linux-mips.org>
Date: Fri, 30 Sep 2005 16:10:52 +0100
Cc: linux-mips@linux-mips.org
In-reply-to: <20050930.123241.72709288.nemoto@toshiba-tops.co.jp>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <20050928.203429.02302175.nemoto@toshiba-tops.co.jp> <20050928.205758.32501424.nemoto@toshiba-tops.co.jp> <20050930.123241.72709288.nemoto@toshiba-tops.co.jp>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.4.2.1i
On Fri, Sep 30, 2005 at 12:32:41PM +0900, Atsushi Nemoto wrote:
> Date: Fri, 30 Sep 2005 12:32:41 +0900 (JST)
> To:   linux-mips@linux-mips.org
> Cc:   ralf@linux-mips.org
> Subject: Re: missing data cache flush for signal trampoline on fork
> From: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
> Content-Type: Text/Plain; charset=us-ascii
> 
> >>>>> On Wed, 28 Sep 2005 20:57:58 +0900 (JST), Atsushi Nemoto 
> >>>>> <anemo@mba.ocn.ne.jp> said:
> anemo> Sorry, this would corrupt cpu_has_ic_fills_f_dc case.  Revised.
> 
> The patch was overkill.  The indexed-flush is required only for
> d-cache.  Revised.

Hmm...  Your patch may be right for the time being but I think this should
the whole flushing biz should actually be handled via update_mmu_cache by
adding something along the lines of:

...
        if (vma->flags & VM_EXEC)
                do_flush_icache_page(addr);
...

What do you think?

  Ralf

<Prev in Thread] Current Thread [Next in Thread>