From mboxrd@z Thu Jan 1 00:00:00 1970 From: Geert Uytterhoeven Date: Wed, 01 Apr 2015 09:37:13 +0000 Subject: [PATCH] mm/migrate: Mark unmap_and_move() "noinline" to avoid ICE in gcc 4.7.3 (was: Re: Possible re Message-Id: List-Id: References: <20150324004537.GA24816@verge.net.au> <20150324161358.GA694@kahuna> <20150326003939.GA25368@verge.net.au> <20150326133631.GB2805@arm.com> <20150327002554.GA5527@verge.net.au> <20150327100612.GB1562@arm.com> <7hbnj99epe.fsf@deeprootsystems.com> <7h8uec95t2.fsf@deeprootsystems.com> In-Reply-To: <7h8uec95t2.fsf@deeprootsystems.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: linux-arm-kernel@lists.infradead.org Hi Kevin, On Tue, 31 Mar 2015, Kevin Hilman wrote: > Ard Biesheuvel writes: > Nope, that branch is already part of linux-next, and linux-next still > fails to compile for 20+ defconfigs[1] >=20 > > Could you elaborate on the issue please? What is the error you are > > getting, and can you confirm that is is caused by ld choking on the > > linker script? If not, this is another error than the one we have been > > trying to fix >=20 > It's definitely not linker script related. >=20 > Using "arm-linux-gnueabi-gcc (Ubuntu/Linaro 4.7.3-12ubuntu1) 4.7.3", > here's the error when building for multi_v7_defconfig (full log > available[2]): >=20 > ../mm/migrate.c: In function 'migrate_pages': > ../mm/migrate.c:1148:1: internal compiler error: in push_minipool_fix, at= config/arm/arm.c:13101 > Please submit a full bug report, > with preprocessed source if appropriate. > See for instructions. > Preprocessed source stored into /tmp/ccO1Nz1m.out file, please attach > this to your bugreport. > make[2]: *** [mm/migrate.o] Error 1 > make[2]: Target `__build' not remade because of errors. > make[1]: *** [mm] Error 2 >=20 > build bisect points to commit 21f992084aeb[3], but that doesn't revert > cleanly so I haven't got any further than that yet. I installed gcc-arm-linux-gnueabi (4:4.7.2-1 from Ubuntu 14.04 LTS) and cou= ld reproduce the ICE. I came up with the workaround below. Does this work for you? >From 7ebe83316eaf1952e55a76754ce7a5832e461b8c Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Wed, 1 Apr 2015 11:22:51 +0200 Subject: [PATCH] mm/migrate: Mark unmap_and_move() "noinline" to avoid ICE = in gcc 4.7.3 MIME-Version: 1.0 Content-Type: text/plain; charset=3DUTF-8 Content-Transfer-Encoding: 8bit With gcc version 4.7.3 (Ubuntu/Linaro 4.7.3-12ubuntu1) : mm/migrate.c: In function =A1migrate_pages=A2: mm/migrate.c:1148:1: internal compiler error: in push_minipool_fix, at = config/arm/arm.c:13500 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. Preprocessed source stored into /tmp/ccPoM1tr.out file, please attach t= his to your bugreport. make[1]: *** [mm/migrate.o] Error 1 make: *** [mm/migrate.o] Error 2 Mark unmap_and_move() (which is used in a single place only) "noinline" to work around this compiler bug. Reported-by: Kevin Hilman Signed-off-by: Geert Uytterhoeven --- mm/migrate.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/mm/migrate.c b/mm/migrate.c index 114602a68111d809..98f8574456c2010c 100644 --- a/mm/migrate.c +++ b/mm/migrate.c @@ -904,9 +904,10 @@ out: * Obtain the lock on page, remove all ptes and migrate the page * to the newly allocated page in newpage. */ -static int unmap_and_move(new_page_t get_new_page, free_page_t put_new_pag= e, - unsigned long private, struct page *page, int force, - enum migrate_mode mode) +static noinline int unmap_and_move(new_page_t get_new_page, + free_page_t put_new_page, + unsigned long private, struct page *page, + int force, enum migrate_mode mode) { int rc =3D 0; int *result =3D NULL; --=20 1.9.1 Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k= .org In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like t= hat. -- Linus Torvalds