Re: [kbuild] [linux-next:master 7884/8197] mm/page_alloc.c:3813 rmqueue_pcplist() warn: inconsistent returns 'flags'.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 7 Jul 2022 09:44:03 +0300 Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:

> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git  master
> head:   088b9c375534d905a4d337c78db3b3bfbb52c4a0
> commit: fdd1b94629bb5592abf40d0c7a572b58e4c0cadc [7884/8197] mm/page_alloc: protect PCP lists with a spinlock
> config: i386-randconfig-m021
> compiler: gcc-11 (Debian 11.3.0-3) 11.3.0
> 
> If you fix the issue, kindly add following tag where applicable
> Reported-by: kernel test robot <lkp@xxxxxxxxx>
> Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> 
> smatch warnings:
> mm/page_alloc.c:3813 rmqueue_pcplist() warn: inconsistent returns 'flags'.
> 
> vim +/flags +3813 mm/page_alloc.c
> 
> 066b23935578d39 Mel Gorman  2017-02-24  3774  static struct page *rmqueue_pcplist(struct zone *preferred_zone,
> 44042b4498728f4 Mel Gorman  2021-06-28  3775  			struct zone *zone, unsigned int order,
> 44042b4498728f4 Mel Gorman  2021-06-28  3776  			gfp_t gfp_flags, int migratetype,
> 44042b4498728f4 Mel Gorman  2021-06-28  3777  			unsigned int alloc_flags)
> 066b23935578d39 Mel Gorman  2017-02-24  3778  {
> 066b23935578d39 Mel Gorman  2017-02-24  3779  	struct per_cpu_pages *pcp;
> 066b23935578d39 Mel Gorman  2017-02-24  3780  	struct list_head *list;
> 066b23935578d39 Mel Gorman  2017-02-24  3781  	struct page *page;
> d34b0733b452ca3 Mel Gorman  2017-04-20  3782  	unsigned long flags;
> fdd1b94629bb559 Mel Gorman  2022-06-24  3783  	unsigned long __maybe_unused UP_flags;
> 066b23935578d39 Mel Gorman  2017-02-24  3784  
> dbbee9d5cd83f9d Mel Gorman  2021-06-28  3785  	local_lock_irqsave(&pagesets.lock, flags);
> 3b12e7e97938424 Mel Gorman  2021-06-28  3786  
> fdd1b94629bb559 Mel Gorman  2022-06-24  3787  	/*
> fdd1b94629bb559 Mel Gorman  2022-06-24  3788  	 * spin_trylock may fail due to a parallel drain. In the future, the
> fdd1b94629bb559 Mel Gorman  2022-06-24  3789  	 * trylock will also protect against IRQ reentrancy.
> fdd1b94629bb559 Mel Gorman  2022-06-24  3790  	 */
> fdd1b94629bb559 Mel Gorman  2022-06-24  3791  	pcp = this_cpu_ptr(zone->per_cpu_pageset);
> fdd1b94629bb559 Mel Gorman  2022-06-24  3792  	pcp_trylock_prepare(UP_flags);
> fdd1b94629bb559 Mel Gorman  2022-06-24  3793  	if (!spin_trylock(&pcp->lock)) {
> fdd1b94629bb559 Mel Gorman  2022-06-24  3794  		pcp_trylock_finish(UP_flags);
> 
> Need to local_unlock_irqrestore(&pagesets.lock, flags); before return

oops.  Looks right.

--- a/mm/page_alloc.c~mm-page_alloc-protect-pcp-lists-with-a-spinlock-fix
+++ a/mm/page_alloc.c
@@ -3792,6 +3792,7 @@ static struct page *rmqueue_pcplist(stru
 	pcp_trylock_prepare(UP_flags);
 	if (!spin_trylock(&pcp->lock)) {
 		pcp_trylock_finish(UP_flags);
+		local_unlock_irqrestore(&pagesets.lock, flags);
 		return NULL;
 	}
 

Do we really need the pcp_trylock_prepare/finish in there? 
local_lock_irqsave() already did that.





[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux