-----Original Message-----
From: Mike Frysinger [mailto:[email protected]]
Sent: 6/30/2010 (星期三) 3:27 上午
To: [email protected]
Cc: [email protected]
Subject: Re: [Linux-kernel-commits] [8952]
trunk/arch/blackfin/include/asm/mmu_context.h: blackfin: restore L1 base
address and length to 0 after free
On Tue, Jun 29, 2010 at 04:55, <[email protected]> wrote:
> blackfin: restore L1 base address and length to 0 after free
>
> Modified: trunk/arch/blackfin/include/asm/mmu_context.h (8951 => 8952)
>
> static inline void free_l1stack(void)
> {
> nr_l1stack_tasks--;
> - if (nr_l1stack_tasks == 0)
>not specific to your change, but this l1stack code looks highly racey.
> shouldnt nr_l1stack-tasks be an atomic_t or something ? and there's
>no locking around the l1_stack_{base,len} ?
>-mike
Agree. There should be a spin_lock to lock the whole to access these global
variants. I'd like to add a new bug thread for this.
_______________________________________________
Linux-kernel-commits mailing list
[email protected]
https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
