From patchwork Thu Oct 8 00:04:26 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Cyril Bur X-Patchwork-Id: 527520 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from lists.ozlabs.org (lists.ozlabs.org [IPv6:2401:3900:2:1::3]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 0787514029C for ; Thu, 8 Oct 2015 11:07:40 +1100 (AEDT) Received: from lists.ozlabs.org (lists.ozlabs.org [IPv6:2401:3900:2:1::3]) by lists.ozlabs.org (Postfix) with ESMTP id E1C931A044F for ; Thu, 8 Oct 2015 11:07:39 +1100 (AEDT) X-Original-To: linuxppc-dev@lists.ozlabs.org Delivered-To: linuxppc-dev@lists.ozlabs.org Received: from ozlabs.org (ozlabs.org [IPv6:2401:3900:2:1::2]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 98C531A022F for ; Thu, 8 Oct 2015 11:05:55 +1100 (AEDT) Received: by ozlabs.org (Postfix) id 77FA4140D95; Thu, 8 Oct 2015 11:05:55 +1100 (AEDT) Delivered-To: linuxppc-dev@ozlabs.org Received: from e23smtp07.au.ibm.com (e23smtp07.au.ibm.com [202.81.31.140]) (using TLSv1 with cipher CAMELLIA256-SHA (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 2AF8D140D93 for ; Thu, 8 Oct 2015 11:05:55 +1100 (AEDT) Received: from /spool/local by e23smtp07.au.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Thu, 8 Oct 2015 10:05:53 +1000 Received: from d23dlp03.au.ibm.com (202.81.31.214) by e23smtp07.au.ibm.com (202.81.31.204) with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted; Thu, 8 Oct 2015 10:05:51 +1000 X-Helo: d23dlp03.au.ibm.com X-MailFrom: cyrilbur@gmail.com X-RcptTo: linuxppc-dev@ozlabs.org Received: from d23relay10.au.ibm.com (d23relay10.au.ibm.com [9.190.26.77]) by d23dlp03.au.ibm.com (Postfix) with ESMTP id 55D12357804F for ; Thu, 8 Oct 2015 11:05:51 +1100 (EST) Received: from d23av02.au.ibm.com (d23av02.au.ibm.com [9.190.235.138]) by d23relay10.au.ibm.com (8.14.9/8.14.9/NCO v10.0) with ESMTP id t9805hDR62455876 for ; Thu, 8 Oct 2015 11:05:51 +1100 Received: from d23av02.au.ibm.com (localhost [127.0.0.1]) by d23av02.au.ibm.com (8.14.4/8.14.4/NCO v10.0 AVout) with ESMTP id t9805IVn021140 for ; Thu, 8 Oct 2015 11:05:19 +1100 Received: from ozlabs.au.ibm.com (ozlabs.au.ibm.com [9.192.253.14]) by d23av02.au.ibm.com (8.14.4/8.14.4/NCO v10.0 AVin) with ESMTP id t9805IYq020761 for ; Thu, 8 Oct 2015 11:05:18 +1100 Received: from camb691.ozlabs.ibm.com (haven.au.ibm.com [9.192.254.114]) (using TLSv1.2 with cipher AES128-SHA256 (128/128 bits)) (No client certificate requested) by ozlabs.au.ibm.com (Postfix) with ESMTPSA id 9EF00A01F9 for ; Thu, 8 Oct 2015 11:04:54 +1100 (AEDT) From: Cyril Bur To: linuxppc-dev@ozlabs.org Subject: [PATCH v2] powerpc: Fix checkstop in native_hpte_clear() with lockdep Date: Thu, 8 Oct 2015 11:04:26 +1100 Message-Id: <1444262666-6942-1-git-send-email-cyrilbur@gmail.com> X-Mailer: git-send-email 2.6.0 X-TM-AS-MML: disable X-Content-Scanned: Fidelis XPS MAILER x-cbid: 15100800-0025-0000-0000-0000023EE759 X-BeenThere: linuxppc-dev@lists.ozlabs.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Errors-To: linuxppc-dev-bounces+patchwork-incoming=ozlabs.org@lists.ozlabs.org Sender: "Linuxppc-dev" native_hpte_clear() is called in real mode from two places: - Early in boot during htab initialisation if firmware assisted dump is active. - Late in the kexec path. In both contexts there is no need to disable interrupts are they are already disabled. Furthermore, locking around the tlbie() is only required for pre POWER5 hardware. On POWER5 or newer hardware concurrent tlbie()s work as expected and on pre POWER5 hardware concurrent tlbie()s could result in deadlock. This code would only be executed at crashdump time, during which all bets are off, concurrent tlbie()s are unlikely and taking locks is unsafe therefore the best course of action is to simply do nothing. Concurrent tlbie()s are not possible in the first case as secondary CPUs have not come up yet. Signed-off-by: Cyril Bur --- Tested on POWER8 system by applying this patch to the petitboot kernel, kexecing into Linus' tree with this patch applied and from there kexecing into Ubuntu 3.19.0-26. v2: No code change. Addition of comment in machdep.h and phrasing tweaks elsewhere. arch/powerpc/include/asm/machdep.h | 9 +++++++-- arch/powerpc/mm/hash_native_64.c | 23 +++++++++++------------ 2 files changed, 18 insertions(+), 14 deletions(-) diff --git a/arch/powerpc/include/asm/machdep.h b/arch/powerpc/include/asm/machdep.h index cab6753..3f191f5 100644 --- a/arch/powerpc/include/asm/machdep.h +++ b/arch/powerpc/include/asm/machdep.h @@ -61,8 +61,13 @@ struct machdep_calls { unsigned long addr, unsigned char *hpte_slot_array, int psize, int ssize, int local); - /* special for kexec, to be called in real mode, linear mapping is - * destroyed as well */ + /* + * Special for kexec. + * To be called in real mode with interrupts disabled. No locks are + * taken as such, concurrent access on pre POWER5 hardware could result + * in a deadlock. + * The linear mapping is destroyed as well. + */ void (*hpte_clear_all)(void); void __iomem * (*ioremap)(phys_addr_t addr, unsigned long size, diff --git a/arch/powerpc/mm/hash_native_64.c b/arch/powerpc/mm/hash_native_64.c index 13befa35..c8822af 100644 --- a/arch/powerpc/mm/hash_native_64.c +++ b/arch/powerpc/mm/hash_native_64.c @@ -582,13 +582,21 @@ static void hpte_decode(struct hash_pte *hpte, unsigned long slot, * be when they isi), and we are the only one left. We rely on our kernel * mapping being 0xC0's and the hardware ignoring those two real bits. * + * This must be called with interrupts disabled. + * + * Taking the native_tlbie_lock is unsafe here due to the possibility of + * lockdep being on. On pre POWER5 hardware, not taking the lock could + * cause deadlock. POWER5 and newer not taking the lock is fine. This only + * gets called during boot before secondary CPUs have come up and during + * crashdump and all bets are off anyway. + * * TODO: add batching support when enabled. remember, no dynamic memory here, * athough there is the control page available... */ static void native_hpte_clear(void) { unsigned long vpn = 0; - unsigned long slot, slots, flags; + unsigned long slot, slots; struct hash_pte *hptep = htab_address; unsigned long hpte_v; unsigned long pteg_count; @@ -596,13 +604,6 @@ static void native_hpte_clear(void) pteg_count = htab_hash_mask + 1; - local_irq_save(flags); - - /* we take the tlbie lock and hold it. Some hardware will - * deadlock if we try to tlbie from two processors at once. - */ - raw_spin_lock(&native_tlbie_lock); - slots = pteg_count * HPTES_PER_GROUP; for (slot = 0; slot < slots; slot++, hptep++) { @@ -614,8 +615,8 @@ static void native_hpte_clear(void) hpte_v = be64_to_cpu(hptep->v); /* - * Call __tlbie() here rather than tlbie() since we - * already hold the native_tlbie_lock. + * Call __tlbie() here rather than tlbie() since we can't take the + * native_tlbie_lock. */ if (hpte_v & HPTE_V_VALID) { hpte_decode(hptep, slot, &psize, &apsize, &ssize, &vpn); @@ -625,8 +626,6 @@ static void native_hpte_clear(void) } asm volatile("eieio; tlbsync; ptesync":::"memory"); - raw_spin_unlock(&native_tlbie_lock); - local_irq_restore(flags); } /*