| Message ID | 20180711231822.GG14131@decadent.org.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Jul 12, 2018 at 12:18:22AM +0100, Ben Hutchings wrote: > The xen-privcmd driver, which can be modular, calls set_pte_at() > which in turn may call __sync_icache_dcache(). > > The call to __sync_icache_dcache() may be optimised out because it is > conditional on !pte_special(), and xen-privcmd calls pte_mkspecial(). > But it seems unwise to rely on this optimisation. > > Fixes: 3ad0876554ca ("xen/privcmd: add IOCTL_PRIVCMD_MMAP_RESOURCE") > Signed-off-by: Ben Hutchings <ben@decadent.org.uk> > --- > This is an RFC since I haven't found an actual build failure. Using > Debian's gcc 7.3 with either CC_OPTIMIZE_FOR_PERFORMANCE or > CC_OPTIMIZE_FOR_SIZE, the call *is* optimised away. Since apply_to_page_range() is EXPORT_SYMBOL_GPL() and a pte_fn_t function would likely use set_pte_at(), I'm fine with this patch for consistency. Acked-by: Catalin Marinas <catalin.marinas@arm.com>
On Fri, Jul 27, 2018 at 01:06:33PM +0100, Catalin Marinas wrote: > On Thu, Jul 12, 2018 at 12:18:22AM +0100, Ben Hutchings wrote: > > The xen-privcmd driver, which can be modular, calls set_pte_at() > > which in turn may call __sync_icache_dcache(). > > > > The call to __sync_icache_dcache() may be optimised out because it is > > conditional on !pte_special(), and xen-privcmd calls pte_mkspecial(). > > But it seems unwise to rely on this optimisation. > > > > Fixes: 3ad0876554ca ("xen/privcmd: add IOCTL_PRIVCMD_MMAP_RESOURCE") > > Signed-off-by: Ben Hutchings <ben@decadent.org.uk> > > --- > > This is an RFC since I haven't found an actual build failure. Using > > Debian's gcc 7.3 with either CC_OPTIMIZE_FOR_PERFORMANCE or > > CC_OPTIMIZE_FOR_SIZE, the call *is* optimised away. > > Since apply_to_page_range() is EXPORT_SYMBOL_GPL() and a > pte_fn_t function would likely use set_pte_at(), I'm fine with this > patch for consistency. > > Acked-by: Catalin Marinas <catalin.marinas@arm.com> Okey doke, I'll pick this up for 4.19. Will
On Fri, Jul 27, 2018 at 3:23 PM Will Deacon <will.deacon@arm.com> wrote: > On Fri, Jul 27, 2018 at 01:06:33PM +0100, Catalin Marinas wrote: > > On Thu, Jul 12, 2018 at 12:18:22AM +0100, Ben Hutchings wrote: > > > The xen-privcmd driver, which can be modular, calls set_pte_at() > > > which in turn may call __sync_icache_dcache(). > > > > > > The call to __sync_icache_dcache() may be optimised out because it is > > > conditional on !pte_special(), and xen-privcmd calls pte_mkspecial(). > > > But it seems unwise to rely on this optimisation. > > > > > > Fixes: 3ad0876554ca ("xen/privcmd: add IOCTL_PRIVCMD_MMAP_RESOURCE") > > > Signed-off-by: Ben Hutchings <ben@decadent.org.uk> > > > --- > > > This is an RFC since I haven't found an actual build failure. Using > > > Debian's gcc 7.3 with either CC_OPTIMIZE_FOR_PERFORMANCE or > > > CC_OPTIMIZE_FOR_SIZE, the call *is* optimised away. > > > > Since apply_to_page_range() is EXPORT_SYMBOL_GPL() and a > > pte_fn_t function would likely use set_pte_at(), I'm fine with this > > patch for consistency. > > > > Acked-by: Catalin Marinas <catalin.marinas@arm.com> > > Okey doke, I'll pick this up for 4.19. Note that this is a build regression in v4.18-rc4 and later: http://kisskb.ellerman.id.au/kisskb/buildresult/13450058/ Gr{oetje,eeting}s, Geert
On Mon, Jul 30, 2018 at 8:46 AM Geert Uytterhoeven <geert@linux-m68k.org> wrote: > On Fri, Jul 27, 2018 at 3:23 PM Will Deacon <will.deacon@arm.com> wrote: > > On Fri, Jul 27, 2018 at 01:06:33PM +0100, Catalin Marinas wrote: > > > On Thu, Jul 12, 2018 at 12:18:22AM +0100, Ben Hutchings wrote: > > > > The xen-privcmd driver, which can be modular, calls set_pte_at() > > > > which in turn may call __sync_icache_dcache(). > > > > > > > > The call to __sync_icache_dcache() may be optimised out because it is > > > > conditional on !pte_special(), and xen-privcmd calls pte_mkspecial(). > > > > But it seems unwise to rely on this optimisation. > > > > > > > > Fixes: 3ad0876554ca ("xen/privcmd: add IOCTL_PRIVCMD_MMAP_RESOURCE") > > > > Signed-off-by: Ben Hutchings <ben@decadent.org.uk> > > > > --- > > > > This is an RFC since I haven't found an actual build failure. Using > > > > Debian's gcc 7.3 with either CC_OPTIMIZE_FOR_PERFORMANCE or > > > > CC_OPTIMIZE_FOR_SIZE, the call *is* optimised away. > > > > > > Since apply_to_page_range() is EXPORT_SYMBOL_GPL() and a > > > pte_fn_t function would likely use set_pte_at(), I'm fine with this > > > patch for consistency. > > > > > > Acked-by: Catalin Marinas <catalin.marinas@arm.com> > > > > Okey doke, I'll pick this up for 4.19. > > Note that this is a build regression in v4.18-rc4 and later: > http://kisskb.ellerman.id.au/kisskb/buildresult/13450058/ s/is/fixes (presumably)/ Gr{oetje,eeting}s, Geert
diff --git a/arch/arm64/mm/flush.c b/arch/arm64/mm/flush.c index 1059884f9a6f..a76f0b983294 100644 --- a/arch/arm64/mm/flush.c +++ b/arch/arm64/mm/flush.c @@ -66,6 +66,7 @@ void __sync_icache_dcache(pte_t pte) sync_icache_aliases(page_address(page), PAGE_SIZE << compound_order(page)); } +EXPORT_SYMBOL_GPL(__sync_icache_dcache); /* * This function is called when a page has been modified by the kernel. Mark
The xen-privcmd driver, which can be modular, calls set_pte_at() which in turn may call __sync_icache_dcache(). The call to __sync_icache_dcache() may be optimised out because it is conditional on !pte_special(), and xen-privcmd calls pte_mkspecial(). But it seems unwise to rely on this optimisation. Fixes: 3ad0876554ca ("xen/privcmd: add IOCTL_PRIVCMD_MMAP_RESOURCE") Signed-off-by: Ben Hutchings <ben@decadent.org.uk> --- This is an RFC since I haven't found an actual build failure. Using Debian's gcc 7.3 with either CC_OPTIMIZE_FOR_PERFORMANCE or CC_OPTIMIZE_FOR_SIZE, the call *is* optimised away. Ben. arch/arm64/mm/flush.c | 1 + 1 file changed, 1 insertion(+)