Hi Russell, On Tue, Oct 1, 2019 at 2:50 AM Russell King - ARM Linux admin <li...@armlinux.org.uk> wrote: > > On Mon, Sep 30, 2019 at 02:59:25PM +0900, Masahiro Yamada wrote: > > KernelCI reports that bcm2835_defconfig is no longer booting since > > commit ac7c3e4ff401 ("compiler: enable CONFIG_OPTIMIZE_INLINING > > forcibly"): > > > > https://lkml.org/lkml/2019/9/26/825 > > > > I also received a regression report from Nicolas Saenz Julienne: > > > > https://lkml.org/lkml/2019/9/27/263 > > > > This problem has cropped up on arch/arm/config/bcm2835_defconfig > > because it enables CONFIG_CC_OPTIMIZE_FOR_SIZE. The compiler tends > > to prefer not inlining functions with -Os. I was able to reproduce > > it with other boards and defconfig files by manually enabling > > CONFIG_CC_OPTIMIZE_FOR_SIZE. > > > > The __get_user_check() specifically uses r0, r1, r2 registers. > > So, uaccess_save_and_enable() and uaccess_restore() must be inlined > > in order to avoid those registers being overwritten in the callees. > > > > Prior to commit 9012d011660e ("compiler: allow all arches to enable > > CONFIG_OPTIMIZE_INLINING"), the 'inline' marker was always enough for > > inlining functions, except on x86. > > > > Since that commit, all architectures can enable CONFIG_OPTIMIZE_INLINING. > > So, __always_inline is now the only guaranteed way of forcible inlining. > > > > I want to keep as much compiler's freedom as possible about the inlining > > decision. So, I changed the function call order instead of adding > > __always_inline around. > > > > Call uaccess_save_and_enable() before assigning the __p ("r0"), and > > uaccess_restore() after evacuating the __e ("r0"). > > > > Fixes: 9012d011660e ("compiler: allow all arches to enable > > CONFIG_OPTIMIZE_INLINING") > > Reported-by: "kernelci.org bot" <b...@kernelci.org> > > Reported-by: Nicolas Saenz Julienne <nsaenzjulie...@suse.de> > > Signed-off-by: Masahiro Yamada <yamada.masah...@socionext.com> > > --- > > > > arch/arm/include/asm/uaccess.h | 8 +++++--- > > 1 file changed, 5 insertions(+), 3 deletions(-) > > > > diff --git a/arch/arm/include/asm/uaccess.h b/arch/arm/include/asm/uaccess.h > > index 303248e5b990..559f252d7e3c 100644 > > --- a/arch/arm/include/asm/uaccess.h > > +++ b/arch/arm/include/asm/uaccess.h > > @@ -191,11 +191,12 @@ extern int __get_user_64t_4(void *); > > #define __get_user_check(x, p) > > \ > > ({ \ > > unsigned long __limit = current_thread_info()->addr_limit - > > 1; \ > > + unsigned int __ua_flags = uaccess_save_and_enable(); \ > > If the compiler is moving uaccess_save_and_enable(), that's something > we really don't want
Hmm, based on my poor knowledge about compilers, I do not know if this re-arrangement happens... > - the idea is to _minimise_ the number of kernel > memory accesses between enabling userspace access and performing the > actual access. > > Fixing it in this way widens the window for the kernel to be doing > something it shoulding in userspace. > > So, the right solution is to ensure that the compiler always inlines > the uaccess_*() helpers - which should be nothing more than four > instructions for uaccess_save_and_enable() and two for the > restore. > OK, I will use __always_inline to avoid any potential behavior change. Thanks. -- Best Regards Masahiro Yamada