From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from gate.crashing.org (gate.crashing.org [63.228.1.57]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTP id BA661DDF7C for ; Wed, 6 Feb 2008 07:53:01 +1100 (EST) Subject: Re: [PATCH] update_mmu_cache: Don't dcbst non-readable pages. From: Benjamin Herrenschmidt To: Scott Wood In-Reply-To: <20080205194326.GA13047@loki.buserror.net> References: <20080205194326.GA13047@loki.buserror.net> Content-Type: text/plain Date: Wed, 06 Feb 2008 07:52:38 +1100 Message-Id: <1202244758.7079.72.camel@pasglop> Mime-Version: 1.0 Cc: linuxppc-dev@ozlabs.org, paulus@samba.org Reply-To: benh@kernel.crashing.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, 2008-02-05 at 13:43 -0600, Scott Wood wrote: > Currently, update_mmu_cache will crash if given a no-access PTE. There's no > need to synchronize dcache/icache unless it's an exec mapping -- however, > due to the existence of older glibc versions that execute out of a > read-but-no-exec page, readability is tested instead. > > This assumes no exec-only mappings; if such mappings become supported, they > will need to go through the kmap_atomic() version of dcache/icache > synchronization. > > Signed-off-by: Scott Wood Acked-by: Benjamin Herrenschmidt > --- > This fixes the following bug: > http://ozlabs.org/pipermail/linuxppc-dev/2008-February/051080.html > > arch/powerpc/mm/mem.c | 7 ++++++- > 1 files changed, 6 insertions(+), 1 deletions(-) > > diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c > index e812244..c7d7bd4 100644 > --- a/arch/powerpc/mm/mem.c > +++ b/arch/powerpc/mm/mem.c > @@ -483,7 +483,12 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, > */ > _tlbie(address, 0 /* 8xx doesn't care about PID */); > #endif > - if (!PageReserved(page) > + /* The _PAGE_USER test should really be _PAGE_EXEC, but > + * older glibc versions execute some code from no-exec > + * pages, which for now we are supporting. If exec-only > + * pages are ever implemented, this will have to change. > + */ > + if (!PageReserved(page) && (pte_val(pte) & _PAGE_USER) > && !test_bit(PG_arch_1, &page->flags)) { > if (vma->vm_mm == current->active_mm) { > __flush_dcache_icache((void *) address);