From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) (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 040CE173 for ; Sun, 30 May 2021 00:51:32 +0000 (UTC) Received: by mail.kernel.org (Postfix) with ESMTPSA id 41D0761131 for ; Sun, 30 May 2021 00:51:32 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1622335892; bh=PWNzmVem9VS//uwig1vEDgd9W083PCsm3baFjr0EQlQ=; h=References:In-Reply-To:From:Date:Subject:To:Cc:From; b=dz5ODyQj3jZj9vOkeVa+Q50fMJBAiUfYq8x2cPJi19s/pSdXtJtMmlqH16qJ5n8/y W3GyGlb+MwGXn3jg/78Tqg9ccouo0cTqzCqBN2XD6LJMuF4ltnhSdQk7jPc85sp6Ac qMOtei5Lo6+xfbAlgysc2kXZHv/mqY23blpHI683qZSQ0uKUbjCizEAO3TFWyFI/qi PB9bL2OV0r5SyxBrVvFEoqV3TgHP/azY4VqE6PGsyssRAMVkLNMNo18PxMyxNRIyQo zkdtde9FniQ0TR4M/ljUOnKiD0HnCrPpsyWA8ZBtYcMVU3ddGHsHzPLiNeL5egwe1b L5pNTAT3NXxiw== Received: by mail-lj1-f179.google.com with SMTP id 131so10131402ljj.3 for ; Sat, 29 May 2021 17:51:32 -0700 (PDT) X-Gm-Message-State: AOAM531QLAvzzboBpKeMV3wXcH1gzUYqDqplc3tC4T5LodMemd2KcgzX FYYWB+tNqkYovuiUXEdTENYKVWbZd3Rgt7WmgQ4= X-Google-Smtp-Source: ABdhPJz2oQ+nB/fKMvvqGrO2vZx+W5md+YIZJDzAEE/bZLcrKJ81oJQTDekOEPMm2E0lbAsEqzAbwAmkbFgNTITKBLQ= X-Received: by 2002:a05:651c:502:: with SMTP id o2mr11500461ljp.105.1622335890517; Sat, 29 May 2021 17:51:30 -0700 (PDT) X-Mailing-List: linux-sunxi@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 References: <20210527070903.GA32653@lst.de> In-Reply-To: From: Guo Ren Date: Sun, 30 May 2021 08:51:19 +0800 X-Gmail-Original-Message-ID: Message-ID: Subject: Re: [PATCH V4 2/2] riscv: Use use_asid_allocator flush TLB To: Palmer Dabbelt Cc: Christoph Hellwig , Anup Patel , Arnd Bergmann , linux-riscv , Linux Kernel Mailing List , linux-arch , linux-sunxi@lists.linux.dev, Guo Ren Content-Type: text/plain; charset="UTF-8" On Sun, May 30, 2021 at 7:42 AM wrote: > > On Thu, 27 May 2021 00:09:03 PDT (-0700), Christoph Hellwig wrote: > > On Wed, May 26, 2021 at 05:49:21AM +0000, guoren@kernel.org wrote: > >> From: Guo Ren > >> > >> Use static_branch_unlikely(&use_asid_allocator) to keep the origin > >> tlb flush style, so it's no effect on the existing machine. Here > >> are the optimized functions: > >> - flush_tlb_mm > >> - flush_tlb_page > >> - flush_tlb_range > >> > >> All above are based on the below new implement functions: > >> - __sbi_tlb_flush_range_asid > >> - local_flush_tlb_range_asid > >> > >> These functions are based on ASID instead of previous non-ASID > >> tlb_flush implementation which invalidates more useful tlb > >> entries. > > > > I still think the commit message is incomplete and rather misleading. > > Here is what I'd come up with from reading the patch: > > > > --------- > > Subject: add ASID-based tlbflushing methods > > > > Implement optimized version of the tlb flushing routines for systems > > using ASIDs. These are behind the use_asid_allocator static branch to > > not affect existing systems not using ASIDs. > > --------- > > That seems much better, thanks. > > >> +static inline void local_flush_tlb_range_asid(unsigned long start, > >> + unsigned long size, unsigned long asid) > >> +{ > >> + unsigned long tmp, end = ALIGN(start + size, PAGE_SIZE); > >> + > >> + for (tmp = start & PAGE_MASK; tmp < end; tmp += PAGE_SIZE) { > >> + __asm__ __volatile__ ("sfence.vma %0, %1" > >> + : > >> + : "r" (tmp), "r" (asid) > >> + : "memory"); > >> + tmp += PAGE_SIZE; > >> + } > > > > This double increments tmp. > > > > Also the non-ASID code switches to a global flush once flushing more > > than a single page. It might be worth documenting the tradeoff in the > > code. > > For some reason I thought we'd written this down in the commentary of > teh ISA manual as the suggested huersitic here, but I can't find it so > maybe I'm wrong. If it's actually there it would be good to point that > out, otherwise some documentation seems fine as it'll probably become a > tunable at some point anyway. >From my view: "non-ASID code switches to a global flush all" is a hardware bug fixup, so I just reserved it to prevent breaking other machines. The comment for the "non-ASID code" should be another patch. > > >> +static void __sbi_tlb_flush_range_asid(struct cpumask *cmask, > >> + unsigned long start, > >> + unsigned long size, > >> + unsigned long asid) > >> +{ > > > > I don't think the calling conventions here are optimal. I'd pass > > the mm_struct as the first argument, as we can derive both the cpumask > > and asid from it instead of doing that in the callers. > > > > But more importantly I think the static branch check can be moved deeper > > into the code to avoid a lot of duplication. What do you think of this > > version? > > > > diff --git a/arch/riscv/include/asm/mmu_context.h b/arch/riscv/include/asm/mmu_context.h > > index b0659413a080..7030837adc1a 100644 > > --- a/arch/riscv/include/asm/mmu_context.h > > +++ b/arch/riscv/include/asm/mmu_context.h > > @@ -33,6 +33,8 @@ static inline int init_new_context(struct task_struct *tsk, > > return 0; > > } > > > > +DECLARE_STATIC_KEY_FALSE(use_asid_allocator); > > + > > #include > > > > #endif /* _ASM_RISCV_MMU_CONTEXT_H */ > > diff --git a/arch/riscv/mm/context.c b/arch/riscv/mm/context.c > > index 68aa312fc352..45c1b04b105d 100644 > > --- a/arch/riscv/mm/context.c > > +++ b/arch/riscv/mm/context.c > > @@ -18,7 +18,7 @@ > > > > #ifdef CONFIG_MMU > > > > -static DEFINE_STATIC_KEY_FALSE(use_asid_allocator); > > +DEFINE_STATIC_KEY_FALSE(use_asid_allocator); > > > > static unsigned long asid_bits; > > static unsigned long num_asids; > > diff --git a/arch/riscv/mm/tlbflush.c b/arch/riscv/mm/tlbflush.c > > index 720b443c4528..d8afbb1269d5 100644 > > --- a/arch/riscv/mm/tlbflush.c > > +++ b/arch/riscv/mm/tlbflush.c > > @@ -4,6 +4,33 @@ > > #include > > #include > > #include > > +#include > > + > > +static inline void local_flush_tlb_all_asid(unsigned long asid) > > +{ > > + __asm__ __volatile__ ("sfence.vma x0, %0" > > + : > > + : "r" (asid) > > + : "memory"); > > +} > > + > > +static inline void local_flush_tlb_page_asid(unsigned long addr, > > + unsigned long asid) > > +{ > > + __asm__ __volatile__ ("sfence.vma %0, %1" > > + : > > + : "r" (addr), "r" (asid) > > + : "memory"); > > +} > > + > > +static inline void local_flush_tlb_range_asid(unsigned long start, > > + unsigned long size, unsigned long asid) > > +{ > > + unsigned long addr, end = ALIGN(start + size, PAGE_SIZE); > > + > > + for (addr = start & PAGE_MASK; addr < end; addr += PAGE_SIZE) > > + local_flush_tlb_page_asid(addr, asid); > > +} > > > > void flush_tlb_all(void) > > { > > @@ -12,28 +39,43 @@ void flush_tlb_all(void) > > > > /* > > * This function must not be called with cmask being null. > > - * Kernel may panic if cmask is NULL. > > */ > > -static void __sbi_tlb_flush_range(struct cpumask *cmask, unsigned long start, > > +static void __sbi_tlb_flush_range(struct mm_struct *mm, unsigned long start, > > unsigned long size) > > { > > + struct cpumask *cmask = mm_cpumask(mm); > > struct cpumask hmask; > > unsigned int cpuid; > > + bool broadcast; > > > > if (cpumask_empty(cmask)) > > return; > > > > cpuid = get_cpu(); > > + /* check if the tlbflush needs to be sent to other CPUs */ > > + broadcast = cpumask_any_but(cmask, cpuid) < nr_cpu_ids; > > + if (static_branch_unlikely(&use_asid_allocator)) { > > + unsigned long asid = atomic_long_read(&mm->context.id); > > > > - if (cpumask_any_but(cmask, cpuid) >= nr_cpu_ids) { > > - /* local cpu is the only cpu present in cpumask */ > > - if (size <= PAGE_SIZE) > > + if (broadcast) { > > + riscv_cpuid_to_hartid_mask(cmask, &hmask); > > + sbi_remote_sfence_vma_asid(cpumask_bits(&hmask), > > + start, size, asid); > > + } else if (size != -1) { > > + local_flush_tlb_range_asid(start, size, asid); > > + } else { > > + local_flush_tlb_all_asid(asid); > > + } > > + } else { > > + if (broadcast) { > > + riscv_cpuid_to_hartid_mask(cmask, &hmask); > > + sbi_remote_sfence_vma(cpumask_bits(&hmask), > > + start, size); > > + } else if (size <= PAGE_SIZE) { > > local_flush_tlb_page(start); > > - else > > + } else { > > local_flush_tlb_all(); > > - } else { > > - riscv_cpuid_to_hartid_mask(cmask, &hmask); > > - sbi_remote_sfence_vma(cpumask_bits(&hmask), start, size); > > + } > > } > > > > put_cpu(); > > @@ -41,16 +83,16 @@ static void __sbi_tlb_flush_range(struct cpumask *cmask, unsigned long start, > > > > void flush_tlb_mm(struct mm_struct *mm) > > { > > - __sbi_tlb_flush_range(mm_cpumask(mm), 0, -1); > > + __sbi_tlb_flush_range(mm, 0, -1); > > } > > > > void flush_tlb_page(struct vm_area_struct *vma, unsigned long addr) > > { > > - __sbi_tlb_flush_range(mm_cpumask(vma->vm_mm), addr, PAGE_SIZE); > > + __sbi_tlb_flush_range(vma->vm_mm, addr, PAGE_SIZE); > > } > > > > void flush_tlb_range(struct vm_area_struct *vma, unsigned long start, > > unsigned long end) > > { > > - __sbi_tlb_flush_range(mm_cpumask(vma->vm_mm), start, end - start); > > + __sbi_tlb_flush_range(vma->vm_mm, start, end - start); > > } > > LGTM. I took the first one as IMO they're really distnict issues, LMK > if you want to re-spin this one or if I should just take what's here. -- Best Regards Guo Ren ML: https://lore.kernel.org/linux-csky/ From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-11.4 required=3.0 tests=BAYES_00,DKIMWL_WL_HIGH, DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,INCLUDES_PATCH,MAILING_LIST_MULTI, SPF_HELO_NONE,SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 0E895C47082 for ; Sun, 30 May 2021 00:51:37 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id DE534601FE for ; Sun, 30 May 2021 00:51:36 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229656AbhE3AxL (ORCPT ); Sat, 29 May 2021 20:53:11 -0400 Received: from mail.kernel.org ([198.145.29.99]:58046 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229546AbhE3AxJ (ORCPT ); Sat, 29 May 2021 20:53:09 -0400 Received: by mail.kernel.org (Postfix) with ESMTPSA id 3D5C561107; Sun, 30 May 2021 00:51:32 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1622335892; bh=PWNzmVem9VS//uwig1vEDgd9W083PCsm3baFjr0EQlQ=; h=References:In-Reply-To:From:Date:Subject:To:Cc:From; b=dz5ODyQj3jZj9vOkeVa+Q50fMJBAiUfYq8x2cPJi19s/pSdXtJtMmlqH16qJ5n8/y W3GyGlb+MwGXn3jg/78Tqg9ccouo0cTqzCqBN2XD6LJMuF4ltnhSdQk7jPc85sp6Ac qMOtei5Lo6+xfbAlgysc2kXZHv/mqY23blpHI683qZSQ0uKUbjCizEAO3TFWyFI/qi PB9bL2OV0r5SyxBrVvFEoqV3TgHP/azY4VqE6PGsyssRAMVkLNMNo18PxMyxNRIyQo zkdtde9FniQ0TR4M/ljUOnKiD0HnCrPpsyWA8ZBtYcMVU3ddGHsHzPLiNeL5egwe1b L5pNTAT3NXxiw== Received: by mail-lj1-f171.google.com with SMTP id p20so10105041ljj.8; Sat, 29 May 2021 17:51:32 -0700 (PDT) X-Gm-Message-State: AOAM5321IGKLvZTIs7TWVh3H3zNTGWK3jIZ2QFmd0tKoDCrJwDMEGKhK ooujJA4ofTUgwlOdBtu792ZEZUyQn8g7gVwsfR4= X-Google-Smtp-Source: ABdhPJz2oQ+nB/fKMvvqGrO2vZx+W5md+YIZJDzAEE/bZLcrKJ81oJQTDekOEPMm2E0lbAsEqzAbwAmkbFgNTITKBLQ= X-Received: by 2002:a05:651c:502:: with SMTP id o2mr11500461ljp.105.1622335890517; Sat, 29 May 2021 17:51:30 -0700 (PDT) MIME-Version: 1.0 References: <20210527070903.GA32653@lst.de> In-Reply-To: From: Guo Ren Date: Sun, 30 May 2021 08:51:19 +0800 X-Gmail-Original-Message-ID: Message-ID: Subject: Re: [PATCH V4 2/2] riscv: Use use_asid_allocator flush TLB To: Palmer Dabbelt Cc: Christoph Hellwig , Anup Patel , Arnd Bergmann , linux-riscv , Linux Kernel Mailing List , linux-arch , linux-sunxi@lists.linux.dev, Guo Ren Content-Type: text/plain; charset="UTF-8" Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sun, May 30, 2021 at 7:42 AM wrote: > > On Thu, 27 May 2021 00:09:03 PDT (-0700), Christoph Hellwig wrote: > > On Wed, May 26, 2021 at 05:49:21AM +0000, guoren@kernel.org wrote: > >> From: Guo Ren > >> > >> Use static_branch_unlikely(&use_asid_allocator) to keep the origin > >> tlb flush style, so it's no effect on the existing machine. Here > >> are the optimized functions: > >> - flush_tlb_mm > >> - flush_tlb_page > >> - flush_tlb_range > >> > >> All above are based on the below new implement functions: > >> - __sbi_tlb_flush_range_asid > >> - local_flush_tlb_range_asid > >> > >> These functions are based on ASID instead of previous non-ASID > >> tlb_flush implementation which invalidates more useful tlb > >> entries. > > > > I still think the commit message is incomplete and rather misleading. > > Here is what I'd come up with from reading the patch: > > > > --------- > > Subject: add ASID-based tlbflushing methods > > > > Implement optimized version of the tlb flushing routines for systems > > using ASIDs. These are behind the use_asid_allocator static branch to > > not affect existing systems not using ASIDs. > > --------- > > That seems much better, thanks. > > >> +static inline void local_flush_tlb_range_asid(unsigned long start, > >> + unsigned long size, unsigned long asid) > >> +{ > >> + unsigned long tmp, end = ALIGN(start + size, PAGE_SIZE); > >> + > >> + for (tmp = start & PAGE_MASK; tmp < end; tmp += PAGE_SIZE) { > >> + __asm__ __volatile__ ("sfence.vma %0, %1" > >> + : > >> + : "r" (tmp), "r" (asid) > >> + : "memory"); > >> + tmp += PAGE_SIZE; > >> + } > > > > This double increments tmp. > > > > Also the non-ASID code switches to a global flush once flushing more > > than a single page. It might be worth documenting the tradeoff in the > > code. > > For some reason I thought we'd written this down in the commentary of > teh ISA manual as the suggested huersitic here, but I can't find it so > maybe I'm wrong. If it's actually there it would be good to point that > out, otherwise some documentation seems fine as it'll probably become a > tunable at some point anyway. >From my view: "non-ASID code switches to a global flush all" is a hardware bug fixup, so I just reserved it to prevent breaking other machines. The comment for the "non-ASID code" should be another patch. > > >> +static void __sbi_tlb_flush_range_asid(struct cpumask *cmask, > >> + unsigned long start, > >> + unsigned long size, > >> + unsigned long asid) > >> +{ > > > > I don't think the calling conventions here are optimal. I'd pass > > the mm_struct as the first argument, as we can derive both the cpumask > > and asid from it instead of doing that in the callers. > > > > But more importantly I think the static branch check can be moved deeper > > into the code to avoid a lot of duplication. What do you think of this > > version? > > > > diff --git a/arch/riscv/include/asm/mmu_context.h b/arch/riscv/include/asm/mmu_context.h > > index b0659413a080..7030837adc1a 100644 > > --- a/arch/riscv/include/asm/mmu_context.h > > +++ b/arch/riscv/include/asm/mmu_context.h > > @@ -33,6 +33,8 @@ static inline int init_new_context(struct task_struct *tsk, > > return 0; > > } > > > > +DECLARE_STATIC_KEY_FALSE(use_asid_allocator); > > + > > #include > > > > #endif /* _ASM_RISCV_MMU_CONTEXT_H */ > > diff --git a/arch/riscv/mm/context.c b/arch/riscv/mm/context.c > > index 68aa312fc352..45c1b04b105d 100644 > > --- a/arch/riscv/mm/context.c > > +++ b/arch/riscv/mm/context.c > > @@ -18,7 +18,7 @@ > > > > #ifdef CONFIG_MMU > > > > -static DEFINE_STATIC_KEY_FALSE(use_asid_allocator); > > +DEFINE_STATIC_KEY_FALSE(use_asid_allocator); > > > > static unsigned long asid_bits; > > static unsigned long num_asids; > > diff --git a/arch/riscv/mm/tlbflush.c b/arch/riscv/mm/tlbflush.c > > index 720b443c4528..d8afbb1269d5 100644 > > --- a/arch/riscv/mm/tlbflush.c > > +++ b/arch/riscv/mm/tlbflush.c > > @@ -4,6 +4,33 @@ > > #include > > #include > > #include > > +#include > > + > > +static inline void local_flush_tlb_all_asid(unsigned long asid) > > +{ > > + __asm__ __volatile__ ("sfence.vma x0, %0" > > + : > > + : "r" (asid) > > + : "memory"); > > +} > > + > > +static inline void local_flush_tlb_page_asid(unsigned long addr, > > + unsigned long asid) > > +{ > > + __asm__ __volatile__ ("sfence.vma %0, %1" > > + : > > + : "r" (addr), "r" (asid) > > + : "memory"); > > +} > > + > > +static inline void local_flush_tlb_range_asid(unsigned long start, > > + unsigned long size, unsigned long asid) > > +{ > > + unsigned long addr, end = ALIGN(start + size, PAGE_SIZE); > > + > > + for (addr = start & PAGE_MASK; addr < end; addr += PAGE_SIZE) > > + local_flush_tlb_page_asid(addr, asid); > > +} > > > > void flush_tlb_all(void) > > { > > @@ -12,28 +39,43 @@ void flush_tlb_all(void) > > > > /* > > * This function must not be called with cmask being null. > > - * Kernel may panic if cmask is NULL. > > */ > > -static void __sbi_tlb_flush_range(struct cpumask *cmask, unsigned long start, > > +static void __sbi_tlb_flush_range(struct mm_struct *mm, unsigned long start, > > unsigned long size) > > { > > + struct cpumask *cmask = mm_cpumask(mm); > > struct cpumask hmask; > > unsigned int cpuid; > > + bool broadcast; > > > > if (cpumask_empty(cmask)) > > return; > > > > cpuid = get_cpu(); > > + /* check if the tlbflush needs to be sent to other CPUs */ > > + broadcast = cpumask_any_but(cmask, cpuid) < nr_cpu_ids; > > + if (static_branch_unlikely(&use_asid_allocator)) { > > + unsigned long asid = atomic_long_read(&mm->context.id); > > > > - if (cpumask_any_but(cmask, cpuid) >= nr_cpu_ids) { > > - /* local cpu is the only cpu present in cpumask */ > > - if (size <= PAGE_SIZE) > > + if (broadcast) { > > + riscv_cpuid_to_hartid_mask(cmask, &hmask); > > + sbi_remote_sfence_vma_asid(cpumask_bits(&hmask), > > + start, size, asid); > > + } else if (size != -1) { > > + local_flush_tlb_range_asid(start, size, asid); > > + } else { > > + local_flush_tlb_all_asid(asid); > > + } > > + } else { > > + if (broadcast) { > > + riscv_cpuid_to_hartid_mask(cmask, &hmask); > > + sbi_remote_sfence_vma(cpumask_bits(&hmask), > > + start, size); > > + } else if (size <= PAGE_SIZE) { > > local_flush_tlb_page(start); > > - else > > + } else { > > local_flush_tlb_all(); > > - } else { > > - riscv_cpuid_to_hartid_mask(cmask, &hmask); > > - sbi_remote_sfence_vma(cpumask_bits(&hmask), start, size); > > + } > > } > > > > put_cpu(); > > @@ -41,16 +83,16 @@ static void __sbi_tlb_flush_range(struct cpumask *cmask, unsigned long start, > > > > void flush_tlb_mm(struct mm_struct *mm) > > { > > - __sbi_tlb_flush_range(mm_cpumask(mm), 0, -1); > > + __sbi_tlb_flush_range(mm, 0, -1); > > } > > > > void flush_tlb_page(struct vm_area_struct *vma, unsigned long addr) > > { > > - __sbi_tlb_flush_range(mm_cpumask(vma->vm_mm), addr, PAGE_SIZE); > > + __sbi_tlb_flush_range(vma->vm_mm, addr, PAGE_SIZE); > > } > > > > void flush_tlb_range(struct vm_area_struct *vma, unsigned long start, > > unsigned long end) > > { > > - __sbi_tlb_flush_range(mm_cpumask(vma->vm_mm), start, end - start); > > + __sbi_tlb_flush_range(vma->vm_mm, start, end - start); > > } > > LGTM. I took the first one as IMO they're really distnict issues, LMK > if you want to re-spin this one or if I should just take what's here. -- Best Regards Guo Ren ML: https://lore.kernel.org/linux-csky/ From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-9.4 required=3.0 tests=BAYES_00,DKIMWL_WL_HIGH, DKIM_SIGNED,DKIM_VALID,INCLUDES_PATCH,MAILING_LIST_MULTI,SPF_HELO_NONE, SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id DB18FC4708F for ; Sun, 30 May 2021 00:52:00 +0000 (UTC) Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id 67C01601FE for ; Sun, 30 May 2021 00:52:00 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 67C01601FE Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=kernel.org Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-riscv-bounces+linux-riscv=archiver.kernel.org@lists.infradead.org DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:Cc:To:Subject:Message-ID:Date:From: In-Reply-To:References:MIME-Version:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=6haM/6/GVxmVlfPqlau8zVLETJoQ9W0Ib761a80oYk0=; b=2g6cvViARgoQKN PBGh/9mQMh79xJ2gTdqq/vqVQdnIRm6qqEvF0xj1qNodFajct9DpXQPTU1OALiAXKoDtaGvBWFljL m+nOFMwo5JVz2TyvnmI1hQ3qGp4ISx/Pwbtmad3PX2nE8Ftx4dU6x0qzEenCPRQY4uTjca0oBLkEt AVvR34Aq0lIvhsi2AcQKDizLShUAb6fHTghI29raIdx4TyD0sa3SMtWWmbn6Esb9mlZjRFJsXEIGM WZs7Hh24IgDUp08/TrWKFpxrAFqdnN8MWuxZQeQRtm/RKqgaou2Lc/J8PrXgOBIJK0S9kNU0WcJgg oLxkB8Qyr3A8uNLkGmew==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.94.2 #2 (Red Hat Linux)) id 1ln9g0-008hk3-9L; Sun, 30 May 2021 00:51:36 +0000 Received: from mail.kernel.org ([198.145.29.99]) by bombadil.infradead.org with esmtps (Exim 4.94.2 #2 (Red Hat Linux)) id 1ln9fw-008hjV-UZ for linux-riscv@lists.infradead.org; Sun, 30 May 2021 00:51:34 +0000 Received: by mail.kernel.org (Postfix) with ESMTPSA id 4449D6112F for ; Sun, 30 May 2021 00:51:32 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1622335892; bh=PWNzmVem9VS//uwig1vEDgd9W083PCsm3baFjr0EQlQ=; h=References:In-Reply-To:From:Date:Subject:To:Cc:From; b=dz5ODyQj3jZj9vOkeVa+Q50fMJBAiUfYq8x2cPJi19s/pSdXtJtMmlqH16qJ5n8/y W3GyGlb+MwGXn3jg/78Tqg9ccouo0cTqzCqBN2XD6LJMuF4ltnhSdQk7jPc85sp6Ac qMOtei5Lo6+xfbAlgysc2kXZHv/mqY23blpHI683qZSQ0uKUbjCizEAO3TFWyFI/qi PB9bL2OV0r5SyxBrVvFEoqV3TgHP/azY4VqE6PGsyssRAMVkLNMNo18PxMyxNRIyQo zkdtde9FniQ0TR4M/ljUOnKiD0HnCrPpsyWA8ZBtYcMVU3ddGHsHzPLiNeL5egwe1b L5pNTAT3NXxiw== Received: by mail-lj1-f179.google.com with SMTP id e2so10117765ljk.4 for ; Sat, 29 May 2021 17:51:32 -0700 (PDT) X-Gm-Message-State: AOAM531KP0wDy9xBTeyYwwTe7MrL/fLr5I2AVmCdJgaz8/MQOb8pfo96 TFFABzuTCxHMPepl6PotVZfOpf2wh7Wtwg4YZy0= X-Google-Smtp-Source: ABdhPJz2oQ+nB/fKMvvqGrO2vZx+W5md+YIZJDzAEE/bZLcrKJ81oJQTDekOEPMm2E0lbAsEqzAbwAmkbFgNTITKBLQ= X-Received: by 2002:a05:651c:502:: with SMTP id o2mr11500461ljp.105.1622335890517; Sat, 29 May 2021 17:51:30 -0700 (PDT) MIME-Version: 1.0 References: <20210527070903.GA32653@lst.de> In-Reply-To: From: Guo Ren Date: Sun, 30 May 2021 08:51:19 +0800 X-Gmail-Original-Message-ID: Message-ID: Subject: Re: [PATCH V4 2/2] riscv: Use use_asid_allocator flush TLB To: Palmer Dabbelt Cc: Christoph Hellwig , Anup Patel , Arnd Bergmann , linux-riscv , Linux Kernel Mailing List , linux-arch , linux-sunxi@lists.linux.dev, Guo Ren X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20210529_175133_060045_C4BF9932 X-CRM114-Status: GOOD ( 45.56 ) X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: "linux-riscv" Errors-To: linux-riscv-bounces+linux-riscv=archiver.kernel.org@lists.infradead.org On Sun, May 30, 2021 at 7:42 AM wrote: > > On Thu, 27 May 2021 00:09:03 PDT (-0700), Christoph Hellwig wrote: > > On Wed, May 26, 2021 at 05:49:21AM +0000, guoren@kernel.org wrote: > >> From: Guo Ren > >> > >> Use static_branch_unlikely(&use_asid_allocator) to keep the origin > >> tlb flush style, so it's no effect on the existing machine. Here > >> are the optimized functions: > >> - flush_tlb_mm > >> - flush_tlb_page > >> - flush_tlb_range > >> > >> All above are based on the below new implement functions: > >> - __sbi_tlb_flush_range_asid > >> - local_flush_tlb_range_asid > >> > >> These functions are based on ASID instead of previous non-ASID > >> tlb_flush implementation which invalidates more useful tlb > >> entries. > > > > I still think the commit message is incomplete and rather misleading. > > Here is what I'd come up with from reading the patch: > > > > --------- > > Subject: add ASID-based tlbflushing methods > > > > Implement optimized version of the tlb flushing routines for systems > > using ASIDs. These are behind the use_asid_allocator static branch to > > not affect existing systems not using ASIDs. > > --------- > > That seems much better, thanks. > > >> +static inline void local_flush_tlb_range_asid(unsigned long start, > >> + unsigned long size, unsigned long asid) > >> +{ > >> + unsigned long tmp, end = ALIGN(start + size, PAGE_SIZE); > >> + > >> + for (tmp = start & PAGE_MASK; tmp < end; tmp += PAGE_SIZE) { > >> + __asm__ __volatile__ ("sfence.vma %0, %1" > >> + : > >> + : "r" (tmp), "r" (asid) > >> + : "memory"); > >> + tmp += PAGE_SIZE; > >> + } > > > > This double increments tmp. > > > > Also the non-ASID code switches to a global flush once flushing more > > than a single page. It might be worth documenting the tradeoff in the > > code. > > For some reason I thought we'd written this down in the commentary of > teh ISA manual as the suggested huersitic here, but I can't find it so > maybe I'm wrong. If it's actually there it would be good to point that > out, otherwise some documentation seems fine as it'll probably become a > tunable at some point anyway. >From my view: "non-ASID code switches to a global flush all" is a hardware bug fixup, so I just reserved it to prevent breaking other machines. The comment for the "non-ASID code" should be another patch. > > >> +static void __sbi_tlb_flush_range_asid(struct cpumask *cmask, > >> + unsigned long start, > >> + unsigned long size, > >> + unsigned long asid) > >> +{ > > > > I don't think the calling conventions here are optimal. I'd pass > > the mm_struct as the first argument, as we can derive both the cpumask > > and asid from it instead of doing that in the callers. > > > > But more importantly I think the static branch check can be moved deeper > > into the code to avoid a lot of duplication. What do you think of this > > version? > > > > diff --git a/arch/riscv/include/asm/mmu_context.h b/arch/riscv/include/asm/mmu_context.h > > index b0659413a080..7030837adc1a 100644 > > --- a/arch/riscv/include/asm/mmu_context.h > > +++ b/arch/riscv/include/asm/mmu_context.h > > @@ -33,6 +33,8 @@ static inline int init_new_context(struct task_struct *tsk, > > return 0; > > } > > > > +DECLARE_STATIC_KEY_FALSE(use_asid_allocator); > > + > > #include > > > > #endif /* _ASM_RISCV_MMU_CONTEXT_H */ > > diff --git a/arch/riscv/mm/context.c b/arch/riscv/mm/context.c > > index 68aa312fc352..45c1b04b105d 100644 > > --- a/arch/riscv/mm/context.c > > +++ b/arch/riscv/mm/context.c > > @@ -18,7 +18,7 @@ > > > > #ifdef CONFIG_MMU > > > > -static DEFINE_STATIC_KEY_FALSE(use_asid_allocator); > > +DEFINE_STATIC_KEY_FALSE(use_asid_allocator); > > > > static unsigned long asid_bits; > > static unsigned long num_asids; > > diff --git a/arch/riscv/mm/tlbflush.c b/arch/riscv/mm/tlbflush.c > > index 720b443c4528..d8afbb1269d5 100644 > > --- a/arch/riscv/mm/tlbflush.c > > +++ b/arch/riscv/mm/tlbflush.c > > @@ -4,6 +4,33 @@ > > #include > > #include > > #include > > +#include > > + > > +static inline void local_flush_tlb_all_asid(unsigned long asid) > > +{ > > + __asm__ __volatile__ ("sfence.vma x0, %0" > > + : > > + : "r" (asid) > > + : "memory"); > > +} > > + > > +static inline void local_flush_tlb_page_asid(unsigned long addr, > > + unsigned long asid) > > +{ > > + __asm__ __volatile__ ("sfence.vma %0, %1" > > + : > > + : "r" (addr), "r" (asid) > > + : "memory"); > > +} > > + > > +static inline void local_flush_tlb_range_asid(unsigned long start, > > + unsigned long size, unsigned long asid) > > +{ > > + unsigned long addr, end = ALIGN(start + size, PAGE_SIZE); > > + > > + for (addr = start & PAGE_MASK; addr < end; addr += PAGE_SIZE) > > + local_flush_tlb_page_asid(addr, asid); > > +} > > > > void flush_tlb_all(void) > > { > > @@ -12,28 +39,43 @@ void flush_tlb_all(void) > > > > /* > > * This function must not be called with cmask being null. > > - * Kernel may panic if cmask is NULL. > > */ > > -static void __sbi_tlb_flush_range(struct cpumask *cmask, unsigned long start, > > +static void __sbi_tlb_flush_range(struct mm_struct *mm, unsigned long start, > > unsigned long size) > > { > > + struct cpumask *cmask = mm_cpumask(mm); > > struct cpumask hmask; > > unsigned int cpuid; > > + bool broadcast; > > > > if (cpumask_empty(cmask)) > > return; > > > > cpuid = get_cpu(); > > + /* check if the tlbflush needs to be sent to other CPUs */ > > + broadcast = cpumask_any_but(cmask, cpuid) < nr_cpu_ids; > > + if (static_branch_unlikely(&use_asid_allocator)) { > > + unsigned long asid = atomic_long_read(&mm->context.id); > > > > - if (cpumask_any_but(cmask, cpuid) >= nr_cpu_ids) { > > - /* local cpu is the only cpu present in cpumask */ > > - if (size <= PAGE_SIZE) > > + if (broadcast) { > > + riscv_cpuid_to_hartid_mask(cmask, &hmask); > > + sbi_remote_sfence_vma_asid(cpumask_bits(&hmask), > > + start, size, asid); > > + } else if (size != -1) { > > + local_flush_tlb_range_asid(start, size, asid); > > + } else { > > + local_flush_tlb_all_asid(asid); > > + } > > + } else { > > + if (broadcast) { > > + riscv_cpuid_to_hartid_mask(cmask, &hmask); > > + sbi_remote_sfence_vma(cpumask_bits(&hmask), > > + start, size); > > + } else if (size <= PAGE_SIZE) { > > local_flush_tlb_page(start); > > - else > > + } else { > > local_flush_tlb_all(); > > - } else { > > - riscv_cpuid_to_hartid_mask(cmask, &hmask); > > - sbi_remote_sfence_vma(cpumask_bits(&hmask), start, size); > > + } > > } > > > > put_cpu(); > > @@ -41,16 +83,16 @@ static void __sbi_tlb_flush_range(struct cpumask *cmask, unsigned long start, > > > > void flush_tlb_mm(struct mm_struct *mm) > > { > > - __sbi_tlb_flush_range(mm_cpumask(mm), 0, -1); > > + __sbi_tlb_flush_range(mm, 0, -1); > > } > > > > void flush_tlb_page(struct vm_area_struct *vma, unsigned long addr) > > { > > - __sbi_tlb_flush_range(mm_cpumask(vma->vm_mm), addr, PAGE_SIZE); > > + __sbi_tlb_flush_range(vma->vm_mm, addr, PAGE_SIZE); > > } > > > > void flush_tlb_range(struct vm_area_struct *vma, unsigned long start, > > unsigned long end) > > { > > - __sbi_tlb_flush_range(mm_cpumask(vma->vm_mm), start, end - start); > > + __sbi_tlb_flush_range(vma->vm_mm, start, end - start); > > } > > LGTM. I took the first one as IMO they're really distnict issues, LMK > if you want to re-spin this one or if I should just take what's here. -- Best Regards Guo Ren ML: https://lore.kernel.org/linux-csky/ _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv