[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20200318153702.GA22501@carbon.DHCP.thefacebook.com>
Date: Wed, 18 Mar 2020 08:37:02 -0700
From: Roman Gushchin <guro@...com>
To: Guido Günther <agx@...xcpu.org>
CC: <linux-kernel@...r.kernel.org>, <linux-next@...r.kernel.org>
Subject: Re: linux-next build failure on arm64 with CMA but !NUMA
On Wed, Mar 18, 2020 at 12:41:09PM +0100, Guido Günther wrote:
> Hi Roman,
> 882d60d2ce725381236a158204d7ec13ff19ab25 in linux-next broke compilation
> on arm64 CMA && !NUMA like
>
> CC kernel/irq/irqdesc.o
> mm/hugetlb.c: In function ‘hugetlb_cma_reserve’:
> mm/hugetlb.c:5449:3: error: implicit declaration of function ‘for_each_mem_pfn_range’; did you mean ‘for_each_mem_range’? [-Werror=implicit-function-declaration]
> for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, NULL) {
> ^~~~~~~~~~~~~~~~~~~~~~
> for_each_mem_range
> mm/hugetlb.c:5449:61: error: expected ‘;’ before ‘{’ token
> for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, NULL) {
> ^~
> Reverting that fixes the build for the moment. Would making all of this
> dependent on CMA && NUMA be the right fix?
Hello Guido!
Thank you for the report!
We've been discussing the issue for a couple of days, and I've proposed a fix.
I've just resent it to linux-mm@ (you're in cc), so hopefully it will be merged today.
Thanks!
Powered by blists - more mailing lists