From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx0a-001b2d01.pphosted.com (mx0a-001b2d01.pphosted.com [148.163.156.1]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 429XFJ5QRqzF3D1 for ; Thu, 13 Sep 2018 05:40:36 +1000 (AEST) Received: from pps.filterd (m0098399.ppops.net [127.0.0.1]) by mx0a-001b2d01.pphosted.com (8.16.0.22/8.16.0.22) with SMTP id w8CJY17Y101610 for ; Wed, 12 Sep 2018 15:40:34 -0400 Received: from e31.co.us.ibm.com (e31.co.us.ibm.com [32.97.110.149]) by mx0a-001b2d01.pphosted.com with ESMTP id 2mf8mj1c7b-1 (version=TLSv1.2 cipher=AES256-GCM-SHA384 bits=256 verify=NOT) for ; Wed, 12 Sep 2018 15:40:34 -0400 Received: from localhost by e31.co.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Wed, 12 Sep 2018 13:40:33 -0600 From: Breno Leitao To: linuxppc-dev@lists.ozlabs.org Cc: mikey@neuling.org, paulus@ozlabs.org, gromero@linux.vnet.ibm.com, mpe@ellerman.id.au, ldufour@linux.vnet.ibm.com, Breno Leitao Subject: [RFC PATCH 02/11] powerpc/tm: Reclaim on unavailable exception Date: Wed, 12 Sep 2018 16:40:10 -0300 In-Reply-To: <1536781219-13938-1-git-send-email-leitao@debian.org> References: <1536781219-13938-1-git-send-email-leitao@debian.org> Message-Id: <1536781219-13938-3-git-send-email-leitao@debian.org> List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , If there is a FP/VEC/Altivec touch inside a transaction and the facility is disabled, then a facility unavailable exception is raised and ends up calling {fp,vec,vsx}_unavailable_tm, which was reclaiming and recheckpointing. This is not required anymore, since the checkpointed state was reclaimed in the trap entrace itself, and it will be recheckpointed by restore_tm_state later. With this new patchset, we only reclaim at exception entrance (except at syscall level if the transaction is suspended), and always recheckpoint on restore_tm_state. Adding a WARN_ON() warning if we hit the _unavailable_tm() in suspended mode, i.e, the reclaim was not executed somehow in the trap entrance. Signed-off-by: Breno Leitao --- arch/powerpc/kernel/exceptions-64s.S | 3 +++ arch/powerpc/kernel/traps.c | 16 ++++------------ 2 files changed, 7 insertions(+), 12 deletions(-) diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S index 78aba71a4b2d..4108f3944bdd 100644 --- a/arch/powerpc/kernel/exceptions-64s.S +++ b/arch/powerpc/kernel/exceptions-64s.S @@ -874,6 +874,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_TM) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM 2: /* User process was in a transaction */ bl save_nvgprs + TM_KERNEL_ENTRY(TM_CAUSE_FAC_UNAV) RECONCILE_IRQ_STATE(r10, r11) addi r3,r1,STACK_FRAME_OVERHEAD bl fp_unavailable_tm @@ -1216,6 +1217,7 @@ BEGIN_FTR_SECTION #ifdef CONFIG_PPC_TRANSACTIONAL_MEM 2: /* User process was in a transaction */ bl save_nvgprs + TM_KERNEL_ENTRY(TM_CAUSE_FAC_UNAV) RECONCILE_IRQ_STATE(r10, r11) addi r3,r1,STACK_FRAME_OVERHEAD bl altivec_unavailable_tm @@ -1252,6 +1254,7 @@ BEGIN_FTR_SECTION #ifdef CONFIG_PPC_TRANSACTIONAL_MEM 2: /* User process was in a transaction */ bl save_nvgprs + TM_KERNEL_ENTRY(TM_CAUSE_FAC_UNAV) RECONCILE_IRQ_STATE(r10, r11) addi r3,r1,STACK_FRAME_OVERHEAD bl vsx_unavailable_tm diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c index c85adb858271..b973fdb72826 100644 --- a/arch/powerpc/kernel/traps.c +++ b/arch/powerpc/kernel/traps.c @@ -1749,19 +1749,12 @@ void fp_unavailable_tm(struct pt_regs *regs) * transaction, and probably retry but now with FP enabled. So the * checkpointed FP registers need to be loaded. */ - tm_reclaim_current(TM_CAUSE_FAC_UNAV); + WARN_ON(MSR_TM_SUSPENDED(mfmsr())); /* Reclaim didn't save out any FPRs to transact_fprs. */ /* Enable FP for the task: */ current->thread.load_fp = 1; - /* This loads and recheckpoints the FP registers from - * thread.fpr[]. They will remain in registers after the - * checkpoint so we don't need to reload them after. - * If VMX is in use, the VRs now hold checkpointed values, - * so we don't want to load the VRs from the thread_struct. - */ - tm_recheckpoint(¤t->thread); } void altivec_unavailable_tm(struct pt_regs *regs) @@ -1773,10 +1766,10 @@ void altivec_unavailable_tm(struct pt_regs *regs) TM_DEBUG("Vector Unavailable trap whilst transactional at 0x%lx," "MSR=%lx\n", regs->nip, regs->msr); - tm_reclaim_current(TM_CAUSE_FAC_UNAV); + WARN_ON(MSR_TM_SUSPENDED(mfmsr())); current->thread.load_vec = 1; - tm_recheckpoint(¤t->thread); current->thread.used_vr = 1; + } void vsx_unavailable_tm(struct pt_regs *regs) @@ -1795,12 +1788,11 @@ void vsx_unavailable_tm(struct pt_regs *regs) current->thread.used_vsr = 1; /* This reclaims FP and/or VR regs if they're already enabled */ - tm_reclaim_current(TM_CAUSE_FAC_UNAV); + WARN_ON(MSR_TM_SUSPENDED(mfmsr())); current->thread.load_vec = 1; current->thread.load_fp = 1; - tm_recheckpoint(¤t->thread); } #endif /* CONFIG_PPC_TRANSACTIONAL_MEM */ -- 2.19.0