From patchwork Thu Sep 24 07:16:54 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Benjamin Herrenschmidt X-Patchwork-Id: 34208 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from bilbo.ozlabs.org (localhost [127.0.0.1]) by ozlabs.org (Postfix) with ESMTP id 19907B7DD1 for ; Thu, 24 Sep 2009 17:22:05 +1000 (EST) Received: by ozlabs.org (Postfix) id BA76AB7B69; Thu, 24 Sep 2009 17:21:59 +1000 (EST) Delivered-To: linuxppc-dev@ozlabs.org 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 ESMTPS id 33AD0B7B65 for ; Thu, 24 Sep 2009 17:21:58 +1000 (EST) Received: from [127.0.0.1] (localhost.localdomain [127.0.0.1]) by gate.crashing.org (8.14.1/8.13.8) with ESMTP id n8O7GxpM022471; Thu, 24 Sep 2009 02:17:00 -0500 Subject: [PATCH-stable] powerpc/8xx: Fix regression introduced by cache coherency rewrite From: Benjamin Herrenschmidt To: stable Date: Thu, 24 Sep 2009 17:16:54 +1000 Message-Id: <1253776614.7103.434.camel@pasglop> Mime-Version: 1.0 X-Mailer: Evolution 2.26.1 Cc: linuxppc-dev list , RFeany@mrv.com X-BeenThere: linuxppc-dev@lists.ozlabs.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: linuxppc-dev-bounces+patchwork-incoming=ozlabs.org@lists.ozlabs.org Errors-To: linuxppc-dev-bounces+patchwork-incoming=ozlabs.org@lists.ozlabs.org From: Rex Feany After upgrading to the latest kernel on my mpc875 userspace started running incredibly slow (hours to get to a shell, even!). I tracked it down to commit 8d30c14cab30d405a05f2aaceda1e9ad57800f36, that patch removed a work-around for the 8xx. Adding it back makes my problem go away. Signed-off-by: Rex Feany Signed-off-by: Benjamin Herrenschmidt --- arch/powerpc/mm/pgtable.c | 19 ++++++++++++++++--- 1 files changed, 16 insertions(+), 3 deletions(-) I'm going to push an upstream variant of this patch upstream tomorrow, it will be a bit different due to significant changes in that file and the affected functions. This should apply to v2.6.30 (with fuzz) and v2.6.31 diff --git a/arch/powerpc/mm/pgtable.c b/arch/powerpc/mm/pgtable.c index 627767d..d8e6725 100644 --- a/arch/powerpc/mm/pgtable.c +++ b/arch/powerpc/mm/pgtable.c @@ -30,6 +30,8 @@ #include #include +#include "mmu_decl.h" + static DEFINE_PER_CPU(struct pte_freelist_batch *, pte_freelist_cur); static unsigned long pte_freelist_forced_free; @@ -119,7 +121,7 @@ void pte_free_finish(void) /* * Handle i/d cache flushing, called from set_pte_at() or ptep_set_access_flags() */ -static pte_t do_dcache_icache_coherency(pte_t pte) +static pte_t do_dcache_icache_coherency(pte_t pte, unsigned long addr) { unsigned long pfn = pte_pfn(pte); struct page *page; @@ -128,6 +130,17 @@ static pte_t do_dcache_icache_coherency(pte_t pte) return pte; page = pfn_to_page(pfn); +#ifdef CONFIG_8xx + /* On 8xx, cache control instructions (particularly + * "dcbst" from flush_dcache_icache) fault as write + * operation if there is an unpopulated TLB entry + * for the address in question. To workaround that, + * we invalidate the TLB here, thus avoiding dcbst + * misbehaviour. + */ + _tlbil_va(addr, 0 /* 8xx doesn't care about PID */); +#endif + if (!PageReserved(page) && !test_bit(PG_arch_1, &page->flags)) { pr_devel("do_dcache_icache_coherency... flushing\n"); flush_dcache_icache_page(page); @@ -198,7 +211,7 @@ void set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte */ pte = __pte(pte_val(pte) & ~_PAGE_HPTEFLAGS); if (pte_need_exec_flush(pte, 1)) - pte = do_dcache_icache_coherency(pte); + pte = do_dcache_icache_coherency(pte, addr); /* Perform the setting of the PTE */ __set_pte_at(mm, addr, ptep, pte, 0); @@ -216,7 +229,7 @@ int ptep_set_access_flags(struct vm_area_struct *vma, unsigned long address, { int changed; if (!dirty && pte_need_exec_flush(entry, 0)) - entry = do_dcache_icache_coherency(entry); + entry = do_dcache_icache_coherency(entry, address); changed = !pte_same(*(ptep), entry); if (changed) { if (!(vma->vm_flags & VM_HUGETLB))