From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 3AF941F941 for ; Fri, 21 Jul 2023 16:17:04 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id AECD7C433CA; Fri, 21 Jul 2023 16:17:03 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1689956224; bh=nYXZSMhhnu5iA7+z3PlljJX3Hm5IHfppWYM4tc1QtCY=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=UqBT9JtnYRngPnLoMKxZM3F0Ms5UNGNoUcWMZ6ph/F2QgH+9SuIdrNzwsoSy8jI1H i+tyxLVac8fDeJfkZgD6aVdBcmbBcScUUhz7DlLGdmVAeCoc5FxMKGL7cFjM5XG4Ol kAlmBIYlV+iXSESvufTYUDPJbt+lhecx+ulP+WOM= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Michael Ellerman Subject: [PATCH 6.4 144/292] powerpc/64s: Fix native_hpte_remove() to be irq-safe Date: Fri, 21 Jul 2023 18:04:13 +0200 Message-ID: <20230721160535.074958891@linuxfoundation.org> X-Mailer: git-send-email 2.41.0 In-Reply-To: <20230721160528.800311148@linuxfoundation.org> References: <20230721160528.800311148@linuxfoundation.org> User-Agent: quilt/0.67 Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: Michael Ellerman commit 8bbe9fee5848371d4af101be445303cac8d880c5 upstream. Lockdep warns that the use of the hpte_lock in native_hpte_remove() is not safe against an IRQ coming in: ================================ WARNING: inconsistent lock state 6.4.0-rc2-g0c54f4d30ecc #1 Not tainted -------------------------------- inconsistent {IN-SOFTIRQ-W} -> {SOFTIRQ-ON-W} usage. qemu-system-ppc/93865 [HC0[0]:SC0[0]:HE1:SE1] takes: c0000000021f5180 (hpte_lock){+.?.}-{0:0}, at: native_lock_hpte+0x8/0xd0 {IN-SOFTIRQ-W} state was registered at: lock_acquire+0x134/0x3f0 native_lock_hpte+0x44/0xd0 native_hpte_insert+0xd4/0x2a0 __hash_page_64K+0x218/0x4f0 hash_page_mm+0x464/0x840 do_hash_fault+0x11c/0x260 data_access_common_virt+0x210/0x220 __ip_select_ident+0x140/0x150 ... net_rx_action+0x3bc/0x440 __do_softirq+0x180/0x534 ... sys_sendmmsg+0x34/0x50 system_call_exception+0x128/0x320 system_call_common+0x160/0x2e4 ... Possible unsafe locking scenario: CPU0 ---- lock(hpte_lock); lock(hpte_lock); *** DEADLOCK *** ... Call Trace: dump_stack_lvl+0x98/0xe0 (unreliable) print_usage_bug.part.0+0x250/0x278 mark_lock+0xc9c/0xd30 __lock_acquire+0x440/0x1ca0 lock_acquire+0x134/0x3f0 native_lock_hpte+0x44/0xd0 native_hpte_remove+0xb0/0x190 kvmppc_mmu_map_page+0x650/0x698 [kvm_pr] kvmppc_handle_pagefault+0x534/0x6e8 [kvm_pr] kvmppc_handle_exit_pr+0x6d8/0xe90 [kvm_pr] after_sprg3_load+0x80/0x90 [kvm_pr] kvmppc_vcpu_run_pr+0x108/0x270 [kvm_pr] kvmppc_vcpu_run+0x34/0x48 [kvm] kvm_arch_vcpu_ioctl_run+0x340/0x470 [kvm] kvm_vcpu_ioctl+0x338/0x8b8 [kvm] sys_ioctl+0x7c4/0x13e0 system_call_exception+0x128/0x320 system_call_common+0x160/0x2e4 I suspect kvm_pr is the only caller that doesn't already have IRQs disabled, which is why this hasn't been reported previously. Fix it by disabling IRQs in native_hpte_remove(). Fixes: 35159b5717fa ("powerpc/64s: make HPTE lock and native_tlbie_lock irq-safe") Cc: stable@vger.kernel.org # v6.1+ Signed-off-by: Michael Ellerman Link: https://msgid.link/20230517123033.18430-1-mpe@ellerman.id.au Signed-off-by: Greg Kroah-Hartman --- arch/powerpc/mm/book3s64/hash_native.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) --- a/arch/powerpc/mm/book3s64/hash_native.c +++ b/arch/powerpc/mm/book3s64/hash_native.c @@ -328,10 +328,12 @@ static long native_hpte_insert(unsigned static long native_hpte_remove(unsigned long hpte_group) { + unsigned long hpte_v, flags; struct hash_pte *hptep; int i; int slot_offset; - unsigned long hpte_v; + + local_irq_save(flags); DBG_LOW(" remove(group=%lx)\n", hpte_group); @@ -356,13 +358,16 @@ static long native_hpte_remove(unsigned slot_offset &= 0x7; } - if (i == HPTES_PER_GROUP) - return -1; + if (i == HPTES_PER_GROUP) { + i = -1; + goto out; + } /* Invalidate the hpte. NOTE: this also unlocks it */ release_hpte_lock(); hptep->v = 0; - +out: + local_irq_restore(flags); return i; }