From patchwork Tue Aug 13 15:01:15 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Rob Herring X-Patchwork-Id: 171202 Delivered-To: patch@linaro.org Received: by 2002:a92:d204:0:0:0:0:0 with SMTP id y4csp1018044ily; Tue, 13 Aug 2019 08:01:23 -0700 (PDT) X-Google-Smtp-Source: APXvYqwXPx+vEJH9Vf+1hXOz6gQlyQRzIuOEbtWn9Vp7d68gMzASxP59XFUY9SsKy13vGD9I7Awy X-Received: by 2002:a17:902:f217:: with SMTP id gn23mr2264132plb.21.1565708483151; Tue, 13 Aug 2019 08:01:23 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1565708483; cv=none; d=google.com; s=arc-20160816; b=FXAMcjsDvkqzkq+qw8LSCUxOv0hqFgd6TRWOtLVmy+hQi8BA5674U9ZnJXgSYRdG06 NzkkxSBafJj4x8EgZwbo2XzZWjqoE2LBWXwxTF1AXT3+EqexXBxS3Hl0FrapfEl4bQZb bjGbyUNXJ0dOFfVvSMCda0T+qr/WCSRnK/mjhjqQ/8+nOqaJOXyNpclvC5P7+Ak9CV63 qPxUcMdj9KWBnYYfbHXSKrJuq07DCiq8AVwYrUbCFs8HMwS4qUjnFmZ4iOb3yw7eWxmc bqzfNj37+TNmBf7mTz0QpL2BIUrwcPNQNLbAU4oQ6SjA8aLtFSFeDmaC2s35PD9NjTOl AamA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:cc:list-subscribe :list-help:list-post:list-archive:list-unsubscribe:list-id :precedence:mime-version:message-id:date:subject:to:from :delivered-to; bh=4WIbsHLI2z9k1JXWGrBssJ9T4X+vAmA1grS+ALGIrWM=; b=Lbzhb1Bhy5KGG7B0rHDPmtFxtZZj4hADlPLpW84rhC7/vFH2AuFWXKkbiTw6ypoW4/ LwhqWBGw43cn/9KxLv/ZL7JYMc5z/IrXhQ0nxN4rd6lVjd2PJGgyn0jWtLk+iNxeN+Al mkKWvvYDmuIWvACkjrTmGjJchl9Z3OURuDsqYqGqNuIKr4XOR6zSGAC5Eb4DgBTpiiTl 5cruwIrdz/XdYamPSRGdQ7jDefn5WRFCQ7DXgvQzP9tjiRDNI+pTOey/r+b6d/4d99Mg IR9tjPcbLlVB5Eb/fzD3QCQTOFwJj+8r3ShZSI6QGhlbHGe8TStpeYFjemiWnH/aCi7G Aj6g== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: best guess record for domain of dri-devel-bounces@lists.freedesktop.org designates 2610:10:20:722:a800:ff:fe36:1795 as permitted sender) smtp.mailfrom=dri-devel-bounces@lists.freedesktop.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=kernel.org Return-Path: Received: from gabe.freedesktop.org (gabe.freedesktop.org. [2610:10:20:722:a800:ff:fe36:1795]) by mx.google.com with ESMTPS id a63si60312237pla.348.2019.08.13.08.01.22 (version=TLS1_2 cipher=ECDHE-RSA-CHACHA20-POLY1305 bits=256/256); Tue, 13 Aug 2019 08:01:23 -0700 (PDT) Received-SPF: pass (google.com: best guess record for domain of dri-devel-bounces@lists.freedesktop.org designates 2610:10:20:722:a800:ff:fe36:1795 as permitted sender) client-ip=2610:10:20:722:a800:ff:fe36:1795; Authentication-Results: mx.google.com; spf=pass (google.com: best guess record for domain of dri-devel-bounces@lists.freedesktop.org designates 2610:10:20:722:a800:ff:fe36:1795 as permitted sender) smtp.mailfrom=dri-devel-bounces@lists.freedesktop.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=kernel.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id DCBDC6E0EE; Tue, 13 Aug 2019 15:01:19 +0000 (UTC) X-Original-To: dri-devel@lists.freedesktop.org Delivered-To: dri-devel@lists.freedesktop.org Received: from mail-ot1-f68.google.com (mail-ot1-f68.google.com [209.85.210.68]) by gabe.freedesktop.org (Postfix) with ESMTPS id B2B686E0EE for ; Tue, 13 Aug 2019 15:01:18 +0000 (UTC) Received: by mail-ot1-f68.google.com with SMTP id o101so20693831ota.8 for ; Tue, 13 Aug 2019 08:01:18 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:mime-version :content-transfer-encoding; bh=NYUcJRW4pIGrio3wIwQXdvClkP/CY2ja2vizZKDz//s=; b=rn1Mojrje8MQ6p3hrHc4q661gmZltK0pJwlW5oQlzMdDw88jYs8wrNoqWJJNW7OheG LhXbzspGWeJb/suSQdWX9LgWu/vFpPzBLq+f0uMPoIvfbuvH60C5tP/aE2gfCyjCpHNb 4sjpI4SgbbQjD2RMj9vNG6ot5R5OgnIKq/+BYNN5AhpGrclYYSfpzNcW2DrbF51/53EL m0+GU+WnhXAhhPHcdRZgiTOXnBfIF0mYSaZBjVbmYSdmLKW4g9biEaAVRybudvSr1GBp ZM1YtQNs7Q++KKOH1nsGu1BHLAjXted32b8Fou8uk6Ec+nhgBzU0j0DTanWmy9vPEHec pelQ== X-Gm-Message-State: APjAAAVQn96WE1TYWfcBVnMB7MUvHpoCP/fqviZZVh00bLgQgGlPMOMS wewHzJv3Vsc31PcOWUSDZ55ZxWA= X-Received: by 2002:a02:6d24:: with SMTP id m36mr44537520jac.87.1565708477425; Tue, 13 Aug 2019 08:01:17 -0700 (PDT) Received: from xps15.herring.priv ([64.188.179.254]) by smtp.googlemail.com with ESMTPSA id o2sm9665648iob.64.2019.08.13.08.01.16 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 13 Aug 2019 08:01:16 -0700 (PDT) From: Rob Herring To: dri-devel@lists.freedesktop.org Subject: [PATCH v2] drm/panfrost: Implement per FD address spaces Date: Tue, 13 Aug 2019 09:01:15 -0600 Message-Id: <20190813150115.30338-1-robh@kernel.org> X-Mailer: git-send-email 2.20.1 MIME-Version: 1.0 X-BeenThere: dri-devel@lists.freedesktop.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: Direct Rendering Infrastructure - Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Tomeu Vizoso , David Airlie , Matthew Auld , Steven Price , Alyssa Rosenzweig , Robin Murphy Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" Up until now, a single shared GPU address space was used. This is not ideal as there's no protection between processes and doesn't work for supporting the same GPU/CPU VA feature. Most importantly, this will hopefully mitigate Alyssa's fear of WebGL, whatever that is. Most of the changes here are moving struct drm_mm and struct panfrost_mmu objects from the per device struct to the per FD struct. The critical function is panfrost_mmu_as_get() which handles allocating and switching the h/w address spaces. There's 3 states an AS can be in: free, allocated, and in use. When a job runs, it requests an address space and then marks it not in use when job is complete(but stays assigned). The first time thru, we find a free AS in the alloc_mask and assign the AS to the FD. Then the next time thru, we most likely already have our AS and we just mark it in use with a ref count. We need a ref count because we have multiple job slots. If the job/FD doesn't have an AS assigned and there are no free ones, then we pick an allocated one not in use from our LRU list and switch the AS from the old FD to the new one. Cc: Tomeu Vizoso Cc: David Airlie Cc: Daniel Vetter Cc: Robin Murphy Cc: Steven Price Cc: Alyssa Rosenzweig Signed-off-by: Rob Herring Acked-by: Alyssa Rosenzweig Reviewed-by: Steven Price --- v2: - Fix assignment of color_adjust ptr - Move clearing of job array ptrs to ISR - Add locking between mmu_hw_do_operation() and panfrost_mmu_as_get() - Drop unnecessary panfrost_mmu_enable() when we already assigned an AS - Use as_lru_list instead of file list to iterate thru all active ASs - Rework the AS reset handling drivers/gpu/drm/panfrost/TODO | 4 - drivers/gpu/drm/panfrost/panfrost_device.c | 2 + drivers/gpu/drm/panfrost/panfrost_device.h | 24 ++- drivers/gpu/drm/panfrost/panfrost_drv.c | 31 ++- drivers/gpu/drm/panfrost/panfrost_gem.c | 15 +- drivers/gpu/drm/panfrost/panfrost_gem.h | 3 + drivers/gpu/drm/panfrost/panfrost_job.c | 14 +- drivers/gpu/drm/panfrost/panfrost_mmu.c | 222 +++++++++++++++------ drivers/gpu/drm/panfrost/panfrost_mmu.h | 8 + 9 files changed, 236 insertions(+), 87 deletions(-) diff --git a/drivers/gpu/drm/panfrost/TODO b/drivers/gpu/drm/panfrost/TODO index e7727b292355..536a0d4f8d29 100644 --- a/drivers/gpu/drm/panfrost/TODO +++ b/drivers/gpu/drm/panfrost/TODO @@ -6,10 +6,6 @@ - Bifrost specific feature and issue handling - Coherent DMA support -- Per FD address space support. The h/w supports multiple addresses spaces. - The hard part is handling when more address spaces are needed than what - the h/w provides. - - Support userspace controlled GPU virtual addresses. Needed for Vulkan. (Tomeu) - Compute job support. So called 'compute only' jobs need to be plumbed up to diff --git a/drivers/gpu/drm/panfrost/panfrost_device.c b/drivers/gpu/drm/panfrost/panfrost_device.c index 9814f4ccbd26..4da71bb56c20 100644 --- a/drivers/gpu/drm/panfrost/panfrost_device.c +++ b/drivers/gpu/drm/panfrost/panfrost_device.c @@ -123,8 +123,10 @@ int panfrost_device_init(struct panfrost_device *pfdev) mutex_init(&pfdev->sched_lock); mutex_init(&pfdev->reset_lock); INIT_LIST_HEAD(&pfdev->scheduled_jobs); + INIT_LIST_HEAD(&pfdev->as_lru_list); spin_lock_init(&pfdev->hwaccess_lock); + spin_lock_init(&pfdev->as_lock); err = panfrost_clk_init(pfdev); if (err) { diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h index 4e5641db9c7e..f503c566e99f 100644 --- a/drivers/gpu/drm/panfrost/panfrost_device.h +++ b/drivers/gpu/drm/panfrost/panfrost_device.h @@ -5,6 +5,8 @@ #ifndef __PANFROST_DEVICE_H__ #define __PANFROST_DEVICE_H__ +#include +#include #include #include #include @@ -63,9 +65,6 @@ struct panfrost_device { spinlock_t hwaccess_lock; - struct drm_mm mm; - spinlock_t mm_lock; - void __iomem *iomem; struct clk *clock; struct clk *bus_clock; @@ -74,7 +73,11 @@ struct panfrost_device { struct panfrost_features features; - struct panfrost_mmu *mmu; + spinlock_t as_lock; + unsigned long as_in_use_mask; + unsigned long as_alloc_mask; + struct list_head as_lru_list; + struct panfrost_job_slot *js; struct panfrost_job *jobs[NUM_JOB_SLOTS]; @@ -98,10 +101,23 @@ struct panfrost_device { } devfreq; }; +struct panfrost_mmu { + struct io_pgtable_cfg pgtbl_cfg; + struct io_pgtable_ops *pgtbl_ops; + struct mutex lock; + int as; + atomic_t as_count; + struct list_head list; +}; + struct panfrost_file_priv { struct panfrost_device *pfdev; struct drm_sched_entity sched_entity[NUM_JOB_SLOTS]; + + struct panfrost_mmu mmu; + struct drm_mm mm; + spinlock_t mm_lock; }; static inline struct panfrost_device *to_panfrost_device(struct drm_device *ddev) diff --git a/drivers/gpu/drm/panfrost/panfrost_drv.c b/drivers/gpu/drm/panfrost/panfrost_drv.c index a1352750984c..fc22f7ca1107 100644 --- a/drivers/gpu/drm/panfrost/panfrost_drv.c +++ b/drivers/gpu/drm/panfrost/panfrost_drv.c @@ -403,6 +403,7 @@ static void panfrost_drm_mm_color_adjust(const struct drm_mm_node *node, static int panfrost_open(struct drm_device *dev, struct drm_file *file) { + int ret; struct panfrost_device *pfdev = dev->dev_private; struct panfrost_file_priv *panfrost_priv; @@ -413,7 +414,28 @@ panfrost_open(struct drm_device *dev, struct drm_file *file) panfrost_priv->pfdev = pfdev; file->driver_priv = panfrost_priv; - return panfrost_job_open(panfrost_priv); + spin_lock_init(&panfrost_priv->mm_lock); + + /* 4G enough for now. can be 48-bit */ + drm_mm_init(&panfrost_priv->mm, SZ_32M >> PAGE_SHIFT, (SZ_4G - SZ_32M) >> PAGE_SHIFT); + panfrost_priv->mm.color_adjust = panfrost_drm_mm_color_adjust; + + ret = panfrost_mmu_pgtable_alloc(panfrost_priv); + if (ret) + goto err_pgtable; + + ret = panfrost_job_open(panfrost_priv); + if (ret) + goto err_job; + + return 0; + +err_job: + panfrost_mmu_pgtable_free(panfrost_priv); +err_pgtable: + drm_mm_takedown(&panfrost_priv->mm); + kfree(panfrost_priv); + return ret; } static void @@ -424,6 +446,8 @@ panfrost_postclose(struct drm_device *dev, struct drm_file *file) panfrost_perfcnt_close(panfrost_priv); panfrost_job_close(panfrost_priv); + panfrost_mmu_pgtable_free(panfrost_priv); + drm_mm_takedown(&panfrost_priv->mm); kfree(panfrost_priv); } @@ -496,14 +520,9 @@ static int panfrost_probe(struct platform_device *pdev) ddev->dev_private = pfdev; pfdev->ddev = ddev; - spin_lock_init(&pfdev->mm_lock); mutex_init(&pfdev->shrinker_lock); INIT_LIST_HEAD(&pfdev->shrinker_list); - /* 4G enough for now. can be 48-bit */ - drm_mm_init(&pfdev->mm, SZ_32M >> PAGE_SHIFT, (SZ_4G - SZ_32M) >> PAGE_SHIFT); - pfdev->mm.color_adjust = panfrost_drm_mm_color_adjust; - pm_runtime_use_autosuspend(pfdev->dev); pm_runtime_set_autosuspend_delay(pfdev->dev, 50); /* ~3 frames */ pm_runtime_enable(pfdev->dev); diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.c b/drivers/gpu/drm/panfrost/panfrost_gem.c index e71f27c4041e..e084bc4e9083 100644 --- a/drivers/gpu/drm/panfrost/panfrost_gem.c +++ b/drivers/gpu/drm/panfrost/panfrost_gem.c @@ -47,8 +47,8 @@ static int panfrost_gem_open(struct drm_gem_object *obj, struct drm_file *file_p size_t size = obj->size; u64 align; struct panfrost_gem_object *bo = to_panfrost_bo(obj); - struct panfrost_device *pfdev = obj->dev->dev_private; unsigned long color = bo->noexec ? PANFROST_BO_NOEXEC : 0; + struct panfrost_file_priv *priv = file_priv->driver_priv; /* * Executable buffers cannot cross a 16MB boundary as the program @@ -61,8 +61,9 @@ static int panfrost_gem_open(struct drm_gem_object *obj, struct drm_file *file_p else align = size >= SZ_2M ? SZ_2M >> PAGE_SHIFT : 0; - spin_lock(&pfdev->mm_lock); - ret = drm_mm_insert_node_generic(&pfdev->mm, &bo->node, + bo->mmu = &priv->mmu; + spin_lock(&priv->mm_lock); + ret = drm_mm_insert_node_generic(&priv->mm, &bo->node, size >> PAGE_SHIFT, align, color, 0); if (ret) goto out; @@ -73,22 +74,22 @@ static int panfrost_gem_open(struct drm_gem_object *obj, struct drm_file *file_p drm_mm_remove_node(&bo->node); } out: - spin_unlock(&pfdev->mm_lock); + spin_unlock(&priv->mm_lock); return ret; } static void panfrost_gem_close(struct drm_gem_object *obj, struct drm_file *file_priv) { struct panfrost_gem_object *bo = to_panfrost_bo(obj); - struct panfrost_device *pfdev = obj->dev->dev_private; + struct panfrost_file_priv *priv = file_priv->driver_priv; if (bo->is_mapped) panfrost_mmu_unmap(bo); - spin_lock(&pfdev->mm_lock); + spin_lock(&priv->mm_lock); if (drm_mm_node_allocated(&bo->node)) drm_mm_remove_node(&bo->node); - spin_unlock(&pfdev->mm_lock); + spin_unlock(&priv->mm_lock); } static int panfrost_gem_pin(struct drm_gem_object *obj) diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.h b/drivers/gpu/drm/panfrost/panfrost_gem.h index e10f58316915..50920819cc16 100644 --- a/drivers/gpu/drm/panfrost/panfrost_gem.h +++ b/drivers/gpu/drm/panfrost/panfrost_gem.h @@ -7,10 +7,13 @@ #include #include +struct panfrost_mmu; + struct panfrost_gem_object { struct drm_gem_shmem_object base; struct sg_table *sgts; + struct panfrost_mmu *mmu; struct drm_mm_node node; bool is_mapped :1; bool noexec :1; diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c index d567ce98494c..ccfac0e5bdd0 100644 --- a/drivers/gpu/drm/panfrost/panfrost_job.c +++ b/drivers/gpu/drm/panfrost/panfrost_job.c @@ -153,6 +153,8 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js) if (WARN_ON(job_read(pfdev, JS_COMMAND_NEXT(js)))) goto end; + cfg = panfrost_mmu_as_get(pfdev, &job->file_priv->mmu); + panfrost_devfreq_record_transition(pfdev, js); spin_lock_irqsave(&pfdev->hwaccess_lock, flags); @@ -163,8 +165,7 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js) /* start MMU, medium priority, cache clean/flush on end, clean/flush on * start */ - /* TODO: different address spaces */ - cfg = JS_CONFIG_THREAD_PRI(8) | + cfg |= JS_CONFIG_THREAD_PRI(8) | JS_CONFIG_START_FLUSH_CLEAN_INVALIDATE | JS_CONFIG_END_FLUSH_CLEAN_INVALIDATE; @@ -377,8 +378,9 @@ static void panfrost_job_timedout(struct drm_sched_job *sched_job) if (dma_fence_is_signaled(job->done_fence)) return; - dev_err(pfdev->dev, "gpu sched timeout, js=%d, status=0x%x, head=0x%x, tail=0x%x, sched_job=%p", + dev_err(pfdev->dev, "gpu sched timeout, js=%d, config=0x%x, status=0x%x, head=0x%x, tail=0x%x, sched_job=%p", js, + job_read(pfdev, JS_CONFIG(js)), job_read(pfdev, JS_STATUS(js)), job_read(pfdev, JS_HEAD_LO(js)), job_read(pfdev, JS_TAIL_LO(js)), @@ -448,8 +450,12 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data) } if (status & JOB_INT_MASK_DONE(j)) { + struct panfrost_job *job = pfdev->jobs[j]; + + pfdev->jobs[j] = NULL; + panfrost_mmu_as_put(pfdev, &job->file_priv->mmu); panfrost_devfreq_record_transition(pfdev, j); - dma_fence_signal(pfdev->jobs[j]->done_fence); + dma_fence_signal(job->done_fence); } status &= ~mask; diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c index 2ed411f09d80..f22d8f02568d 100644 --- a/drivers/gpu/drm/panfrost/panfrost_mmu.c +++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c @@ -1,5 +1,6 @@ // SPDX-License-Identifier: GPL-2.0 /* Copyright 2019 Linaro, Ltd, Rob Herring */ +#include #include #include #include @@ -22,12 +23,6 @@ #define mmu_write(dev, reg, data) writel(data, dev->iomem + reg) #define mmu_read(dev, reg) readl(dev->iomem + reg) -struct panfrost_mmu { - struct io_pgtable_cfg pgtbl_cfg; - struct io_pgtable_ops *pgtbl_ops; - struct mutex lock; -}; - static int wait_ready(struct panfrost_device *pfdev, u32 as_nr) { int ret; @@ -85,13 +80,19 @@ static void lock_region(struct panfrost_device *pfdev, u32 as_nr, } -static int mmu_hw_do_operation(struct panfrost_device *pfdev, u32 as_nr, - u64 iova, size_t size, u32 op) +static int mmu_hw_do_operation(struct panfrost_device *pfdev, + struct panfrost_mmu *mmu, + u64 iova, size_t size, u32 op) { - unsigned long flags; - int ret; + int ret, as_nr; - spin_lock_irqsave(&pfdev->hwaccess_lock, flags); + spin_lock(&pfdev->as_lock); + as_nr = mmu->as; + + if (as_nr < 0) { + spin_unlock(&pfdev->as_lock); + return 0; + } if (op != AS_COMMAND_UNLOCK) lock_region(pfdev, as_nr, iova, size); @@ -102,14 +103,15 @@ static int mmu_hw_do_operation(struct panfrost_device *pfdev, u32 as_nr, /* Wait for the flush to complete */ ret = wait_ready(pfdev, as_nr); - spin_unlock_irqrestore(&pfdev->hwaccess_lock, flags); + spin_unlock(&pfdev->as_lock); return ret; } -static void panfrost_mmu_enable(struct panfrost_device *pfdev, u32 as_nr) +static void panfrost_mmu_enable(struct panfrost_device *pfdev, struct panfrost_mmu *mmu) { - struct io_pgtable_cfg *cfg = &pfdev->mmu->pgtbl_cfg; + int as_nr = mmu->as; + struct io_pgtable_cfg *cfg = &mmu->pgtbl_cfg; u64 transtab = cfg->arm_mali_lpae_cfg.transtab; u64 memattr = cfg->arm_mali_lpae_cfg.memattr; @@ -136,9 +138,75 @@ static void mmu_disable(struct panfrost_device *pfdev, u32 as_nr) write_cmd(pfdev, as_nr, AS_COMMAND_UPDATE); } +u32 panfrost_mmu_as_get(struct panfrost_device *pfdev, struct panfrost_mmu *mmu) +{ + int as; + + spin_lock(&pfdev->as_lock); + + as = mmu->as; + if (as >= 0) { + int en = atomic_inc_return(&mmu->as_count); + WARN_ON(en >= NUM_JOB_SLOTS); + + list_move(&mmu->list, &pfdev->as_lru_list); + goto out; + } + + /* Check for a free AS */ + as = ffz(pfdev->as_alloc_mask); + if (!(BIT(as) & pfdev->features.as_present)) { + struct panfrost_mmu *lru_mmu; + + list_for_each_entry_reverse(lru_mmu, &pfdev->as_lru_list, list) { + if (!atomic_read(&lru_mmu->as_count)) + break; + } + WARN_ON(&lru_mmu->list == &pfdev->as_lru_list); + + list_del_init(&lru_mmu->list); + as = lru_mmu->as; + + WARN_ON(as < 0); + lru_mmu->as = -1; + } + + /* Assign the free or reclaimed AS to the FD */ + mmu->as = as; + set_bit(as, &pfdev->as_alloc_mask); + atomic_set(&mmu->as_count, 1); + list_add(&mmu->list, &pfdev->as_lru_list); + + dev_dbg(pfdev->dev, "Assigned AS%d to mmu %p, alloc_mask=%lx", as, mmu, pfdev->as_alloc_mask); + + panfrost_mmu_enable(pfdev, mmu); + +out: + spin_unlock(&pfdev->as_lock); + return as; +} + +void panfrost_mmu_as_put(struct panfrost_device *pfdev, struct panfrost_mmu *mmu) +{ + atomic_dec(&mmu->as_count); + WARN_ON(atomic_read(&mmu->as_count) < 0); +} + void panfrost_mmu_reset(struct panfrost_device *pfdev) { - panfrost_mmu_enable(pfdev, 0); + struct panfrost_mmu *mmu, *mmu_tmp; + + spin_lock(&pfdev->as_lock); + + pfdev->as_alloc_mask = 0; + + list_for_each_entry_safe(mmu, mmu_tmp, &pfdev->as_lru_list, list) { + mmu->as = -1; + atomic_set(&mmu->as_count, 0); + list_del_init(&mmu->list); + } + + spin_unlock(&pfdev->as_lock); mmu_write(pfdev, MMU_INT_CLEAR, ~0); mmu_write(pfdev, MMU_INT_MASK, ~0); @@ -152,21 +220,21 @@ static size_t get_pgsize(u64 addr, size_t size) return SZ_2M; } -static int mmu_map_sg(struct panfrost_device *pfdev, u64 iova, - int prot, struct sg_table *sgt) +static int mmu_map_sg(struct panfrost_device *pfdev, struct panfrost_mmu *mmu, + u64 iova, int prot, struct sg_table *sgt) { unsigned int count; struct scatterlist *sgl; - struct io_pgtable_ops *ops = pfdev->mmu->pgtbl_ops; + struct io_pgtable_ops *ops = mmu->pgtbl_ops; u64 start_iova = iova; - mutex_lock(&pfdev->mmu->lock); + mutex_lock(&mmu->lock); for_each_sg(sgt->sgl, sgl, sgt->nents, count) { unsigned long paddr = sg_dma_address(sgl); size_t len = sg_dma_len(sgl); - dev_dbg(pfdev->dev, "map: iova=%llx, paddr=%lx, len=%zx", iova, paddr, len); + dev_dbg(pfdev->dev, "map: as=%d, iova=%llx, paddr=%lx, len=%zx", mmu->as, iova, paddr, len); while (len) { size_t pgsize = get_pgsize(iova | paddr, len); @@ -178,10 +246,10 @@ static int mmu_map_sg(struct panfrost_device *pfdev, u64 iova, } } - mmu_hw_do_operation(pfdev, 0, start_iova, iova - start_iova, + mmu_hw_do_operation(pfdev, mmu, start_iova, iova - start_iova, AS_COMMAND_FLUSH_PT); - mutex_unlock(&pfdev->mmu->lock); + mutex_unlock(&mmu->lock); return 0; } @@ -208,7 +276,7 @@ int panfrost_mmu_map(struct panfrost_gem_object *bo) if (ret < 0) return ret; - mmu_map_sg(pfdev, bo->node.start << PAGE_SHIFT, prot, sgt); + mmu_map_sg(pfdev, bo->mmu, bo->node.start << PAGE_SHIFT, prot, sgt); pm_runtime_mark_last_busy(pfdev->dev); pm_runtime_put_autosuspend(pfdev->dev); @@ -221,7 +289,7 @@ void panfrost_mmu_unmap(struct panfrost_gem_object *bo) { struct drm_gem_object *obj = &bo->base.base; struct panfrost_device *pfdev = to_panfrost_device(obj->dev); - struct io_pgtable_ops *ops = pfdev->mmu->pgtbl_ops; + struct io_pgtable_ops *ops = bo->mmu->pgtbl_ops; u64 iova = bo->node.start << PAGE_SHIFT; size_t len = bo->node.size << PAGE_SHIFT; size_t unmapped_len = 0; @@ -230,13 +298,13 @@ void panfrost_mmu_unmap(struct panfrost_gem_object *bo) if (WARN_ON(!bo->is_mapped)) return; - dev_dbg(pfdev->dev, "unmap: iova=%llx, len=%zx", iova, len); + dev_dbg(pfdev->dev, "unmap: as=%d, iova=%llx, len=%zx", bo->mmu->as, iova, len); ret = pm_runtime_get_sync(pfdev->dev); if (ret < 0) return; - mutex_lock(&pfdev->mmu->lock); + mutex_lock(&bo->mmu->lock); while (unmapped_len < len) { size_t unmapped_page; @@ -250,10 +318,10 @@ void panfrost_mmu_unmap(struct panfrost_gem_object *bo) unmapped_len += pgsize; } - mmu_hw_do_operation(pfdev, 0, bo->node.start << PAGE_SHIFT, + mmu_hw_do_operation(pfdev, bo->mmu, bo->node.start << PAGE_SHIFT, bo->node.size << PAGE_SHIFT, AS_COMMAND_FLUSH_PT); - mutex_unlock(&pfdev->mmu->lock); + mutex_unlock(&bo->mmu->lock); pm_runtime_mark_last_busy(pfdev->dev); pm_runtime_put_autosuspend(pfdev->dev); @@ -262,9 +330,9 @@ void panfrost_mmu_unmap(struct panfrost_gem_object *bo) static void mmu_tlb_inv_context_s1(void *cookie) { - struct panfrost_device *pfdev = cookie; + struct panfrost_file_priv *priv = cookie; - mmu_hw_do_operation(pfdev, 0, 0, ~0UL, AS_COMMAND_FLUSH_MEM); + mmu_hw_do_operation(priv->pfdev, &priv->mmu, 0, ~0UL, AS_COMMAND_FLUSH_MEM); } static void mmu_tlb_inv_range_nosync(unsigned long iova, size_t size, @@ -283,16 +351,69 @@ static const struct iommu_gather_ops mmu_tlb_ops = { .tlb_sync = mmu_tlb_sync_context, }; +int panfrost_mmu_pgtable_alloc(struct panfrost_file_priv *priv) +{ + struct panfrost_mmu *mmu = &priv->mmu; + struct panfrost_device *pfdev = priv->pfdev; + + mutex_init(&mmu->lock); + INIT_LIST_HEAD(&mmu->list); + mmu->as = -1; + + mmu->pgtbl_cfg = (struct io_pgtable_cfg) { + .pgsize_bitmap = SZ_4K | SZ_2M, + .ias = FIELD_GET(0xff, pfdev->features.mmu_features), + .oas = FIELD_GET(0xff00, pfdev->features.mmu_features), + .tlb = &mmu_tlb_ops, + .iommu_dev = pfdev->dev, + }; + + mmu->pgtbl_ops = alloc_io_pgtable_ops(ARM_MALI_LPAE, &mmu->pgtbl_cfg, + priv); + if (!mmu->pgtbl_ops) + return -EINVAL; + + return 0; +} + +void panfrost_mmu_pgtable_free(struct panfrost_file_priv *priv) +{ + struct panfrost_device *pfdev = priv->pfdev; + struct panfrost_mmu *mmu = &priv->mmu; + + spin_lock(&pfdev->as_lock); + if (mmu->as >= 0) { + clear_bit(mmu->as, &pfdev->as_alloc_mask); + clear_bit(mmu->as, &pfdev->as_in_use_mask); + list_del(&mmu->list); + } + spin_unlock(&pfdev->as_lock); + + free_io_pgtable_ops(mmu->pgtbl_ops); +} + static struct drm_mm_node *addr_to_drm_mm_node(struct panfrost_device *pfdev, int as, u64 addr) { - struct drm_mm_node *node; + struct drm_mm_node *node = NULL; u64 offset = addr >> PAGE_SHIFT; + struct panfrost_mmu *mmu; - drm_mm_for_each_node(node, &pfdev->mm) { - if (offset >= node->start && offset < (node->start + node->size)) - return node; + spin_lock(&pfdev->as_lock); + list_for_each_entry(mmu, &pfdev->as_lru_list, list) { + struct panfrost_file_priv *priv; + if (as != mmu->as) + continue; + + priv = container_of(mmu, struct panfrost_file_priv, mmu); + drm_mm_for_each_node(node, &priv->mm) { + if (offset >= node->start && offset < (node->start + node->size)) + goto out; + } } - return NULL; + +out: + spin_unlock(&pfdev->as_lock); + return node; } #define NUM_FAULT_PAGES (SZ_2M / PAGE_SIZE) @@ -317,6 +438,8 @@ int panfrost_mmu_map_fault_addr(struct panfrost_device *pfdev, int as, u64 addr) node->start << PAGE_SHIFT); return -EINVAL; } + WARN_ON(bo->mmu->as != as); + /* Assume 2MB alignment and size multiple */ addr &= ~((u64)SZ_2M - 1); page_offset = addr >> PAGE_SHIFT; @@ -367,11 +490,11 @@ int panfrost_mmu_map_fault_addr(struct panfrost_device *pfdev, int as, u64 addr) goto err_map; } - mmu_map_sg(pfdev, addr, IOMMU_WRITE | IOMMU_READ | IOMMU_NOEXEC, sgt); + mmu_map_sg(pfdev, bo->mmu, addr, IOMMU_WRITE | IOMMU_READ | IOMMU_NOEXEC, sgt); bo->is_mapped = true; - dev_dbg(pfdev->dev, "mapped page fault @ %llx", addr); + dev_dbg(pfdev->dev, "mapped page fault @ AS%d %llx", as, addr); return 0; @@ -480,15 +603,8 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data) int panfrost_mmu_init(struct panfrost_device *pfdev) { - struct io_pgtable_ops *pgtbl_ops; int err, irq; - pfdev->mmu = devm_kzalloc(pfdev->dev, sizeof(*pfdev->mmu), GFP_KERNEL); - if (!pfdev->mmu) - return -ENOMEM; - - mutex_init(&pfdev->mmu->lock); - irq = platform_get_irq_byname(to_platform_device(pfdev->dev), "mmu"); if (irq <= 0) return -ENODEV; @@ -501,22 +617,6 @@ int panfrost_mmu_init(struct panfrost_device *pfdev) dev_err(pfdev->dev, "failed to request mmu irq"); return err; } - pfdev->mmu->pgtbl_cfg = (struct io_pgtable_cfg) { - .pgsize_bitmap = SZ_4K | SZ_2M, - .ias = FIELD_GET(0xff, pfdev->features.mmu_features), - .oas = FIELD_GET(0xff00, pfdev->features.mmu_features), - .tlb = &mmu_tlb_ops, - .iommu_dev = pfdev->dev, - }; - - pgtbl_ops = alloc_io_pgtable_ops(ARM_MALI_LPAE, &pfdev->mmu->pgtbl_cfg, - pfdev); - if (!pgtbl_ops) - return -ENOMEM; - - pfdev->mmu->pgtbl_ops = pgtbl_ops; - - panfrost_mmu_enable(pfdev, 0); return 0; } @@ -525,6 +625,4 @@ void panfrost_mmu_fini(struct panfrost_device *pfdev) { mmu_write(pfdev, MMU_INT_MASK, 0); mmu_disable(pfdev, 0); - - free_io_pgtable_ops(pfdev->mmu->pgtbl_ops); } diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.h b/drivers/gpu/drm/panfrost/panfrost_mmu.h index d5f9b24537db..7c5b6775ae23 100644 --- a/drivers/gpu/drm/panfrost/panfrost_mmu.h +++ b/drivers/gpu/drm/panfrost/panfrost_mmu.h @@ -5,6 +5,8 @@ #define __PANFROST_MMU_H__ struct panfrost_gem_object; +struct panfrost_file_priv; +struct panfrost_mmu; int panfrost_mmu_map(struct panfrost_gem_object *bo); void panfrost_mmu_unmap(struct panfrost_gem_object *bo); @@ -13,4 +15,10 @@ int panfrost_mmu_init(struct panfrost_device *pfdev); void panfrost_mmu_fini(struct panfrost_device *pfdev); void panfrost_mmu_reset(struct panfrost_device *pfdev); +u32 panfrost_mmu_as_get(struct panfrost_device *pfdev, struct panfrost_mmu *mmu); +void panfrost_mmu_as_put(struct panfrost_device *pfdev, struct panfrost_mmu *mmu); + +int panfrost_mmu_pgtable_alloc(struct panfrost_file_priv *priv); +void panfrost_mmu_pgtable_free(struct panfrost_file_priv *priv); + #endif