From patchwork Mon Oct 21 19:27:43 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Matthew Auld X-Patchwork-Id: 11202947 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 5950B14ED for ; Mon, 21 Oct 2019 19:28:04 +0000 (UTC) Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (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 41E2A205ED for ; Mon, 21 Oct 2019 19:28:04 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 41E2A205ED Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 2B9646E22F; Mon, 21 Oct 2019 19:28:03 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mga03.intel.com (mga03.intel.com [134.134.136.65]) by gabe.freedesktop.org (Postfix) with ESMTPS id B49396E22B for ; Mon, 21 Oct 2019 19:27:52 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga002.fm.intel.com ([10.253.24.26]) by orsmga103.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 21 Oct 2019 12:27:52 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.67,324,1566889200"; d="scan'208";a="227432333" Received: from cmclare-mobl.ger.corp.intel.com (HELO mwahaha-bdw.ger.corp.intel.com) ([10.252.7.98]) by fmsmga002.fm.intel.com with ESMTP; 21 Oct 2019 12:27:50 -0700 From: Matthew Auld To: intel-gfx@lists.freedesktop.org Date: Mon, 21 Oct 2019 20:27:43 +0100 Message-Id: <20191021192747.24804-3-matthew.auld@intel.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20191021192747.24804-1-matthew.auld@intel.com> References: <20191021192747.24804-1-matthew.auld@intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH v2 3/7] drm/i915/lmem: support kernel mapping 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" From: Abdiel Janulgue We can create LMEM objects, but we also need to support mapping them into kernel space for internal use. Signed-off-by: Abdiel Janulgue Signed-off-by: Matthew Auld Signed-off-by: Steve Hampson Cc: Joonas Lahtinen Reviewed-by: Chris Wilson --- drivers/gpu/drm/i915/gem/i915_gem_lmem.c | 36 ++++++ drivers/gpu/drm/i915/gem/i915_gem_lmem.h | 8 ++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 +- drivers/gpu/drm/i915/gem/i915_gem_pages.c | 22 +++- .../drm/i915/selftests/intel_memory_region.c | 113 ++++++++++++++++++ 5 files changed, 180 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_lmem.c b/drivers/gpu/drm/i915/gem/i915_gem_lmem.c index 5168ab9fa4ce..919ce0736c69 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_lmem.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_lmem.c @@ -9,11 +9,47 @@ #include "i915_drv.h" const struct drm_i915_gem_object_ops i915_gem_lmem_obj_ops = { + .flags = I915_GEM_OBJECT_HAS_IOMEM, + .get_pages = i915_gem_object_get_pages_buddy, .put_pages = i915_gem_object_put_pages_buddy, .release = i915_gem_object_release_memory_region, }; +/* XXX: Time to vfunc your life up? */ +void __iomem *i915_gem_object_lmem_io_map_page(struct drm_i915_gem_object *obj, + unsigned long n) +{ + resource_size_t offset; + + offset = i915_gem_object_get_dma_address(obj, n); + + return io_mapping_map_wc(&obj->mm.region->iomap, offset, PAGE_SIZE); +} + +void __iomem *i915_gem_object_lmem_io_map_page_atomic(struct drm_i915_gem_object *obj, + unsigned long n) +{ + resource_size_t offset; + + offset = i915_gem_object_get_dma_address(obj, n); + + return io_mapping_map_atomic_wc(&obj->mm.region->iomap, offset); +} + +void __iomem *i915_gem_object_lmem_io_map(struct drm_i915_gem_object *obj, + unsigned long n, + unsigned long size) +{ + resource_size_t offset; + + GEM_BUG_ON(!i915_gem_object_is_contiguous(obj)); + + offset = i915_gem_object_get_dma_address(obj, n); + + return io_mapping_map_wc(&obj->mm.region->iomap, offset, size); +} + bool i915_gem_object_is_lmem(struct drm_i915_gem_object *obj) { return obj->ops == &i915_gem_lmem_obj_ops; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_lmem.h b/drivers/gpu/drm/i915/gem/i915_gem_lmem.h index fc3f15580fe3..7c176b8b7d2f 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_lmem.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_lmem.h @@ -14,6 +14,14 @@ struct intel_memory_region; extern const struct drm_i915_gem_object_ops i915_gem_lmem_obj_ops; +void __iomem *i915_gem_object_lmem_io_map(struct drm_i915_gem_object *obj, + unsigned long n, unsigned long size); +void __iomem *i915_gem_object_lmem_io_map_page(struct drm_i915_gem_object *obj, + unsigned long n); +void __iomem * +i915_gem_object_lmem_io_map_page_atomic(struct drm_i915_gem_object *obj, + unsigned long n); + bool i915_gem_object_is_lmem(struct drm_i915_gem_object *obj); struct drm_i915_gem_object * diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h index a387e3ee728b..96008374a412 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h @@ -31,10 +31,11 @@ struct i915_lut_handle { struct drm_i915_gem_object_ops { unsigned int flags; #define I915_GEM_OBJECT_HAS_STRUCT_PAGE BIT(0) -#define I915_GEM_OBJECT_IS_SHRINKABLE BIT(1) -#define I915_GEM_OBJECT_IS_PROXY BIT(2) -#define I915_GEM_OBJECT_NO_GGTT BIT(3) -#define I915_GEM_OBJECT_ASYNC_CANCEL BIT(4) +#define I915_GEM_OBJECT_HAS_IOMEM BIT(1) +#define I915_GEM_OBJECT_IS_SHRINKABLE BIT(2) +#define I915_GEM_OBJECT_IS_PROXY BIT(3) +#define I915_GEM_OBJECT_NO_GGTT BIT(4) +#define I915_GEM_OBJECT_ASYNC_CANCEL BIT(5) /* Interface between the GEM object and its backing storage. * get_pages() is called once prior to the use of the associated set diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pages.c b/drivers/gpu/drm/i915/gem/i915_gem_pages.c index b0ec0959c13f..cf7f5a3cb210 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_pages.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_pages.c @@ -7,6 +7,7 @@ #include "i915_drv.h" #include "i915_gem_object.h" #include "i915_scatterlist.h" +#include "i915_gem_lmem.h" void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj, struct sg_table *pages, @@ -172,7 +173,9 @@ __i915_gem_object_unset_pages(struct drm_i915_gem_object *obj) void *ptr; ptr = page_mask_bits(obj->mm.mapping); - if (is_vmalloc_addr(ptr)) + if (i915_gem_object_is_lmem(obj)) + io_mapping_unmap(ptr); + else if (is_vmalloc_addr(ptr)) vunmap(ptr); else kunmap(kmap_to_page(ptr)); @@ -231,7 +234,7 @@ int __i915_gem_object_put_pages(struct drm_i915_gem_object *obj, } /* The 'mapping' part of i915_gem_object_pin_map() below */ -static void *i915_gem_object_map(const struct drm_i915_gem_object *obj, +static void *i915_gem_object_map(struct drm_i915_gem_object *obj, enum i915_map_type type) { unsigned long n_pages = obj->base.size >> PAGE_SHIFT; @@ -244,6 +247,13 @@ static void *i915_gem_object_map(const struct drm_i915_gem_object *obj, pgprot_t pgprot; void *addr; + if (i915_gem_object_is_lmem(obj)) { + if (type != I915_MAP_WC) + return NULL; + + return i915_gem_object_lmem_io_map(obj, 0, obj->base.size); + } + /* A single page can always be kmapped */ if (n_pages == 1 && type == I915_MAP_WB) return kmap(sg_page(sgt->sgl)); @@ -285,11 +295,13 @@ void *i915_gem_object_pin_map(struct drm_i915_gem_object *obj, enum i915_map_type type) { enum i915_map_type has_type; + unsigned int flags; bool pinned; void *ptr; int err; - if (unlikely(!i915_gem_object_has_struct_page(obj))) + flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE | I915_GEM_OBJECT_HAS_IOMEM; + if (!i915_gem_object_type_has(obj, flags)) return ERR_PTR(-ENXIO); err = mutex_lock_interruptible(&obj->mm.lock); @@ -321,7 +333,9 @@ void *i915_gem_object_pin_map(struct drm_i915_gem_object *obj, goto err_unpin; } - if (is_vmalloc_addr(ptr)) + if (i915_gem_object_is_lmem(obj)) + io_mapping_unmap(ptr); + else if (is_vmalloc_addr(ptr)) vunmap(ptr); else kunmap(kmap_to_page(ptr)); diff --git a/drivers/gpu/drm/i915/selftests/intel_memory_region.c b/drivers/gpu/drm/i915/selftests/intel_memory_region.c index 617a35cfac2f..292489371842 100644 --- a/drivers/gpu/drm/i915/selftests/intel_memory_region.c +++ b/drivers/gpu/drm/i915/selftests/intel_memory_region.c @@ -13,8 +13,10 @@ #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "gem/i915_gem_object_blt.h" #include "gem/selftests/mock_context.h" #include "gt/intel_gt.h" +#include "selftests/igt_flush_test.h" #include "selftests/i915_random.h" static void close_objects(struct intel_memory_region *mem, @@ -275,6 +277,116 @@ static int igt_lmem_create(void *arg) return err; } +static int igt_lmem_write_cpu(void *arg) +{ + struct drm_i915_private *i915 = arg; + struct drm_i915_gem_object *obj; + I915_RND_STATE(prng); + IGT_TIMEOUT(end_time); + u32 bytes[] = { + 0, /* rng placeholder */ + sizeof(u32), + sizeof(u64), + 64, /* cl */ + PAGE_SIZE, + PAGE_SIZE - sizeof(u32), + PAGE_SIZE - sizeof(u64), + PAGE_SIZE - 64, + }; + u32 *vaddr; + u32 sz; + u32 i; + int *order; + int count; + int err; + + if (!HAS_ENGINE(i915, BCS0)) + return 0; + + sz = round_up(prandom_u32_state(&prng) % SZ_32M, PAGE_SIZE); + sz = max_t(u32, 2 * PAGE_SIZE, sz); + + obj = i915_gem_object_create_lmem(i915, sz, I915_BO_ALLOC_CONTIGUOUS); + if (IS_ERR(obj)) + return PTR_ERR(obj); + + vaddr = i915_gem_object_pin_map(obj, I915_MAP_WC); + if (IS_ERR(vaddr)) { + err = PTR_ERR(vaddr); + goto out_put; + } + + /* Put the pages into a known state -- do so from the gpu for added fun */ + err = i915_gem_object_fill_blt(obj, i915->engine[BCS0]->kernel_context, + 0xdeadbeaf); + if (err) + goto out_unpin; + + i915_gem_object_lock(obj); + err = i915_gem_object_set_to_wc_domain(obj, true); + i915_gem_object_unlock(obj); + if (err) + goto out_unpin; + + count = ARRAY_SIZE(bytes); + order = i915_random_order(count * count, &prng); + if (!order) { + err = -ENOMEM; + goto out_unpin; + } + + /* We want to throw in a random width/align */ + bytes[0] = igt_random_offset(&prng, 0, PAGE_SIZE, sizeof(u32), + sizeof(u32)); + + i = 0; + do { + u32 offset; + u32 align; + u32 dword; + u32 size; + u32 val; + + size = bytes[order[i] % count]; + i = (i + 1) % (count * count); + + align = bytes[order[i] % count]; + i = (i + 1) % (count * count); + + align = max_t(u32, sizeof(u32), rounddown_pow_of_two(align)); + + offset = igt_random_offset(&prng, 0, obj->base.size, + size, align); + + val = prandom_u32_state(&prng); + memset32(vaddr + offset / sizeof(u32), val ^ 0xdeadbeaf, + size / sizeof(u32)); + + /* + * Sample random dw -- don't waste precious time reading every + * single dw. + */ + dword = igt_random_offset(&prng, offset, + offset + size, + sizeof(u32), sizeof(u32)); + dword /= sizeof(u32); + if (vaddr[dword] != (val ^ 0xdeadbeaf)) { + pr_err("%s vaddr[%u]=%u, val=%u, size=%u, align=%u, offset=%u\n", + __func__, dword, vaddr[dword], val ^ 0xdeadbeaf, + size, align, offset); + err = -EINVAL; + break; + } + } while (!__igt_timeout(end_time, NULL)); + +out_unpin: + i915_gem_object_unpin_map(obj); +out_put: + i915_gem_object_put(obj); + + return err; +} + int intel_memory_region_mock_selftests(void) { static const struct i915_subtest tests[] = { @@ -308,6 +420,7 @@ int intel_memory_region_live_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { SUBTEST(igt_lmem_create), + SUBTEST(igt_lmem_write_cpu), }; if (!HAS_LMEM(i915)) {