From patchwork Fri Jul 5 21:52:04 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Chris Wilson X-Patchwork-Id: 11033387 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id CDD891890 for ; Fri, 5 Jul 2019 21:52:13 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id BDFE427FB0 for ; Fri, 5 Jul 2019 21:52:13 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id B223228B01; Fri, 5 Jul 2019 21:52:13 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-5.2 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher DHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id 3049928ABC for ; Fri, 5 Jul 2019 21:52:13 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 465646E55F; Fri, 5 Jul 2019 21:52:10 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from fireflyinternet.com (mail.fireflyinternet.com [109.228.58.192]) by gabe.freedesktop.org (Postfix) with ESMTPS id 9E32D6E55E for ; Fri, 5 Jul 2019 21:52:08 +0000 (UTC) X-Default-Received-SPF: pass (skip=forwardok (res=PASS)) x-ip-name=78.156.65.138; Received: from haswell.alporthouse.com (unverified [78.156.65.138]) by fireflyinternet.com (Firefly Internet (M1)) with ESMTP id 17148575-1500050 for ; Fri, 05 Jul 2019 22:52:05 +0100 From: Chris Wilson To: intel-gfx@lists.freedesktop.org Date: Fri, 5 Jul 2019 22:52:04 +0100 Message-Id: <20190705215204.4559-4-chris@chris-wilson.co.uk> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20190705215204.4559-1-chris@chris-wilson.co.uk> References: <20190705215204.4559-1-chris@chris-wilson.co.uk> MIME-Version: 1.0 Subject: [Intel-gfx] [CI 4/4] drm/i915/gtt: Introduce release_pd_entry X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" X-Virus-Scanned: ClamAV using ClamSMTP From: Mika Kuoppala By encapsulating the locking upper level and used check for entry into a helper function, we can use it in all callsites. v2: get rid of atomic_reads on lower level clears (Chris) Cc: Chris Wilson Signed-off-by: Mika Kuoppala Reviewed-by: Chris Wilson --- drivers/gpu/drm/i915/i915_gem_gtt.c | 114 ++++++++++------------------ 1 file changed, 42 insertions(+), 72 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c index 0a55b0932c86..236c964dd761 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -811,7 +811,25 @@ __clear_pd_entry(struct i915_page_directory * const pd, __set_pd_entry((pd), (pde), px_base(to), gen8_pde_encode) #define clear_pd_entry(pd, pde, to) \ - __clear_pd_entry((pd), (pde), px_base(to), gen8_pde_encode) + __clear_pd_entry((pd), (pde), (to), gen8_pde_encode) + +static bool +release_pd_entry(struct i915_page_directory * const pd, + const unsigned short pde, + atomic_t *counter, + struct i915_page_dma * const scratch) +{ + bool free = false; + + spin_lock(&pd->lock); + if (atomic_dec_and_test(counter)) { + clear_pd_entry(pd, pde, scratch); + free = true; + } + spin_unlock(&pd->lock); + + return free; +} /* * PDE TLBs are a pain to invalidate on GEN8+. When we modify @@ -827,11 +845,11 @@ static void mark_tlbs_dirty(struct i915_ppgtt *ppgtt) /* Removes entries from a single page table, releasing it if it's empty. * Caller can use the return value to update higher-level entries. */ -static bool gen8_ppgtt_clear_pt(const struct i915_address_space *vm, +static void gen8_ppgtt_clear_pt(const struct i915_address_space *vm, struct i915_page_table *pt, u64 start, u64 length) { - unsigned int num_entries = gen8_pte_count(start, length); + const unsigned int num_entries = gen8_pte_count(start, length); gen8_pte_t *vaddr; vaddr = kmap_atomic_px(pt); @@ -839,10 +857,11 @@ static bool gen8_ppgtt_clear_pt(const struct i915_address_space *vm, kunmap_atomic(vaddr); GEM_BUG_ON(num_entries > atomic_read(&pt->used)); - return !atomic_sub_return(num_entries, &pt->used); + + atomic_sub(num_entries, &pt->used); } -static bool gen8_ppgtt_clear_pd(struct i915_address_space *vm, +static void gen8_ppgtt_clear_pd(struct i915_address_space *vm, struct i915_page_directory *pd, u64 start, u64 length) { @@ -850,30 +869,20 @@ static bool gen8_ppgtt_clear_pd(struct i915_address_space *vm, u32 pde; gen8_for_each_pde(pt, pd, start, length, pde) { - bool free = false; - GEM_BUG_ON(pt == vm->scratch_pt); - if (!gen8_ppgtt_clear_pt(vm, pt, start, length)) - continue; - - spin_lock(&pd->lock); - if (!atomic_read(&pt->used)) { - clear_pd_entry(pd, pde, vm->scratch_pt); - free = true; - } - spin_unlock(&pd->lock); - if (free) + atomic_inc(&pt->used); + gen8_ppgtt_clear_pt(vm, pt, start, length); + if (release_pd_entry(pd, pde, &pt->used, + px_base(vm->scratch_pt))) free_pt(vm, pt); } - - return !atomic_read(&pd->used); } /* Removes entries from a single page dir pointer, releasing it if it's empty. * Caller can use the return value to update higher-level entries */ -static bool gen8_ppgtt_clear_pdp(struct i915_address_space *vm, +static void gen8_ppgtt_clear_pdp(struct i915_address_space *vm, struct i915_page_directory * const pdp, u64 start, u64 length) { @@ -881,24 +890,14 @@ static bool gen8_ppgtt_clear_pdp(struct i915_address_space *vm, unsigned int pdpe; gen8_for_each_pdpe(pd, pdp, start, length, pdpe) { - bool free = false; - GEM_BUG_ON(pd == vm->scratch_pd); - if (!gen8_ppgtt_clear_pd(vm, pd, start, length)) - continue; - - spin_lock(&pdp->lock); - if (!atomic_read(&pd->used)) { - clear_pd_entry(pdp, pdpe, vm->scratch_pd); - free = true; - } - spin_unlock(&pdp->lock); - if (free) + atomic_inc(&pd->used); + gen8_ppgtt_clear_pd(vm, pd, start, length); + if (release_pd_entry(pdp, pdpe, &pd->used, + px_base(vm->scratch_pd))) free_pd(vm, pd); } - - return !atomic_read(&pdp->used); } static void gen8_ppgtt_clear_3lvl(struct i915_address_space *vm, @@ -922,19 +921,12 @@ static void gen8_ppgtt_clear_4lvl(struct i915_address_space *vm, GEM_BUG_ON(!i915_vm_is_4lvl(vm)); gen8_for_each_pml4e(pdp, pml4, start, length, pml4e) { - bool free = false; GEM_BUG_ON(pdp == vm->scratch_pdp); - if (!gen8_ppgtt_clear_pdp(vm, pdp, start, length)) - continue; - - spin_lock(&pml4->lock); - if (!atomic_read(&pdp->used)) { - clear_pd_entry(pml4, pml4e, vm->scratch_pdp); - free = true; - } - spin_unlock(&pml4->lock); - if (free) + atomic_inc(&pdp->used); + gen8_ppgtt_clear_pdp(vm, pdp, start, length); + if (release_pd_entry(pml4, pml4e, &pdp->used, + px_base(vm->scratch_pdp))) free_pd(vm, pdp); } } @@ -1457,17 +1449,8 @@ static int gen8_ppgtt_alloc_pdp(struct i915_address_space *vm, goto out; unwind_pd: - if (alloc) { - free_pd(vm, alloc); - alloc = NULL; - } - spin_lock(&pdp->lock); - if (atomic_dec_and_test(&pd->used)) { - GEM_BUG_ON(alloc); - alloc = pd; /* defer the free to after the lock */ - clear_pd_entry(pdp, pdpe, vm->scratch_pd); - } - spin_unlock(&pdp->lock); + if (release_pd_entry(pdp, pdpe, &pd->used, px_base(vm->scratch_pd))) + free_pd(vm, pd); unwind: gen8_ppgtt_clear_pdp(vm, pdp, from, start - from); out: @@ -1530,17 +1513,8 @@ static int gen8_ppgtt_alloc_4lvl(struct i915_address_space *vm, goto out; unwind_pdp: - if (alloc) { - free_pd(vm, alloc); - alloc = NULL; - } - spin_lock(&pml4->lock); - if (atomic_dec_and_test(&pdp->used)) { - GEM_BUG_ON(alloc); - alloc = pdp; /* defer the free until after the lock */ - clear_pd_entry(pml4, pml4e, vm->scratch_pdp); - } - spin_unlock(&pml4->lock); + if (release_pd_entry(pml4, pml4e, &pdp->used, px_base(vm->scratch_pdp))) + free_pd(vm, pdp); unwind: gen8_ppgtt_clear_4lvl(vm, from, start - from); out: @@ -1570,11 +1544,7 @@ static int gen8_preallocate_top_level_pdp(struct i915_ppgtt *ppgtt) return 0; unwind: - start -= from; - gen8_for_each_pdpe(pd, pdp, from, start, pdpe) { - clear_pd_entry(pdp, pdpe, vm->scratch_pd); - free_pd(vm, pd); - } + gen8_ppgtt_clear_pdp(vm, pdp, from, start - from); atomic_set(&pdp->used, 0); return -ENOMEM; }