<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <style type="text/css" style="display:none;"></style> </head> <body dir="ltr"> <div id="divtagdefaultwrapper" style="color: rgb(0, 0, 0); font-family: Calibri, Helvetica, sans-serif, EmojiFont, "Apple Color Emoji", "Segoe UI Emoji", NotoColorEmoji, "Segoe UI Symbol", "Android Emoji", EmojiSymbols;" dir="ltr"> Some minor comments as below. Reviewed-by: Bruce Chang <yu.bruce.chang@intel.com> <div style="color: rgb(0, 0, 0);"> <div style="font-size: 12pt;"> <hr tabindex="-1" style="display:inline-block; width:98%"> <div id="x_divRplyFwdMsg" dir="ltr">From: Chris Wilson <chris@chris-wilson.co.uk> Sent: Saturday, December 12, 2020 1:43 AM To: intel-gfx@lists.freedesktop.org Cc: igt-dev@lists.freedesktop.org; Chris Wilson; Chang, Yu bruce Subject: [PATCH i-g-t] lib: Pass device fd to gem_mmappable_aperture_size() <div> </div> </div> </div> <div class="PlainText" style="font-size: 10pt;">In order to find the correct aperture size for the test, we want to pass the test's device into the query. Reported-by: Bruce Chang <yu.bruce.chang@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Bruce Chang <yu.bruce.chang@intel.com> --- lib/i915/gem_mman.c | 108 +++++++++++++++++++++++++++ lib/i915/gem_mman.h | 6 ++ lib/ioctl_wrappers.c | 116 ----------------------------- lib/ioctl_wrappers.h | 5 -- tests/i915/gem_concurrent_all.c | 12 +-- tests/i915/gem_mmap.c | 4 +- tests/i915/gem_mmap_gtt.c | 10 +-- tests/i915/gem_pwrite.c | 6 +- tests/i915/gem_shrink.c | 2 +- tests/i915/gem_tiled_fence_blits.c | 2 +- tests/i915/i915_pm_rpm.c | 4 +- tests/kms_big_fb.c | 2 +- tests/kms_flip.c | 2 +- tests/prime_mmap.c | 4 +- 14 files changed, 138 insertions(+), 145 deletions(-) diff --git a/lib/i915/gem_mman.c b/lib/i915/gem_mman.c index 93bef2bfc..703c6a5c1 100644 --- a/lib/i915/gem_mman.c +++ b/lib/i915/gem_mman.c @@ -27,7 +27,9 @@ #include <errno.h> #include "igt_core.h" +#include "igt_device.h" #include "ioctl_wrappers.h" +#include "intel_chipset.h" #include "gem_mman.h" @@ -551,3 +553,109 @@ const struct mmap_offset mmap_offset_types[] = { { "uc", I915_MMAP_OFFSET_UC, I915_GEM_DOMAIN_WC }, {}, }; + +/** + * gem_available_aperture_size: + * @fd: open i915 drm file descriptor + * + * Feature test macro to query the kernel for the available gpu aperture size + * usable in a batchbuffer. + * + * Returns: The available gtt address space size. + */ +uint64_t gem_available_aperture_size(int fd) +{ + struct drm_i915_gem_get_aperture aperture = { + aperture.aper_available_size = 256*1024*1024, + }; + + ioctl(fd, DRM_IOCTL_I915_GEM_GET_APERTURE, &aperture); + errno = 0; + + return aperture.aper_available_size; +} + +/** + * gem_aperture_size: + * @fd: open i915 drm file descriptor + * + * Feature test macro to query the kernel for the total gpu aperture size. + * + * Returns: The total gtt address space size. + */ +uint64_t gem_aperture_size(int fd) +{ + struct drm_i915_gem_context_param p = { + .param = 0x3 + }; </div> <div class="PlainText" style="font-size: 10pt;">Understand this is the original code, but it will be more readable to use I915_CONTEXT_PARAM_GTT_SIZE instead of 3.</div> <div class="PlainText" style="font-size: 10pt;"> </div> <div class="PlainText" style="font-size: 10pt;">+ + if (__gem_context_get_param(fd, &p)) + p.value = gem_global_aperture_size(fd); + + return p.value; +} + +/** + * gem_mappable_aperture_size: + * + * Feature test macro to query the kernel for the mappable gpu aperture size. + * This is the area available for GTT memory mappings. + * + * Returns: The mappable gtt address space size. + */ +uint64_t gem_mappable_aperture_size(int fd) +{ + struct pci_device *pci_dev = igt_device_get_pci_device(fd);</div> <div class="PlainText" style="font-size: 10pt;"> </div> <div class="PlainText" style="font-size: 10pt;">Does it make sense to eliminate the function intel_get_pci_device() if not being used anymore? But it can be a separate patch.</div> <div class="PlainText" style=""> + int bar; + + if (intel_gen(pci_dev->device_id) < 3) + bar = 0; + else + bar = 2; + + return pci_dev->regions[bar].size; +} + +/** + * gem_global_aperture_size: + * @fd: open i915 drm file descriptor + * + * Feature test macro to query the kernel for the global gpu aperture size. + * This is the area available for the kernel to perform address translations. + * + * Returns: The gtt address space size. + */ +uint64_t gem_global_aperture_size(int fd) +{ + struct drm_i915_gem_get_aperture aperture = { + aperture.aper_size = 256*1024*1024 + }; + + ioctl(fd, DRM_IOCTL_I915_GEM_GET_APERTURE, &aperture); + errno = 0; + + return aperture.aper_size; +} + +/** + * gem_available_fences: + * @fd: open i915 drm file descriptor + * + * Feature test macro to query the kernel for the number of available fences + * usable in a batchbuffer. Only relevant for pre-gen4. + * + * Returns: The number of available fences. + */ +int gem_available_fences(int fd) +{ + int num_fences = 0; + struct drm_i915_getparam gp = { + gp.param = I915_PARAM_NUM_FENCES_AVAIL, + gp.value = &num_fences, + }; + + ioctl(fd, DRM_IOCTL_I915_GETPARAM, &gp, sizeof(gp)); + errno = 0; + + return num_fences; +} diff --git a/lib/i915/gem_mman.h b/lib/i915/gem_mman.h index 2c4a7a00b..ec2899ffe 100644 --- a/lib/i915/gem_mman.h +++ b/lib/i915/gem_mman.h @@ -109,5 +109,11 @@ bool gem_has_mmap_offset_type(int fd, const struct mmap_offset *t); (__t)++) \ for_each_if(gem_has_mmap_offset_type((fd), (__t))) +uint64_t gem_available_aperture_size(int fd); +uint64_t gem_aperture_size(int fd); +uint64_t gem_global_aperture_size(int fd); +uint64_t gem_mappable_aperture_size(int fd); +int gem_available_fences(int fd); + #endif /* GEM_MMAN_H */ diff --git a/lib/ioctl_wrappers.c b/lib/ioctl_wrappers.c index 06431ce6c..a928f894f 100644 --- a/lib/ioctl_wrappers.c +++ b/lib/ioctl_wrappers.c @@ -782,31 +782,6 @@ bool gem_engine_reset_enabled(int fd) return gem_gpu_reset_type(fd) > 1; } -/** - * gem_available_fences: - * @fd: open i915 drm file descriptor - * - * Feature test macro to query the kernel for the number of available fences - * usable in a batchbuffer. Only relevant for pre-gen4. - * - * Returns: The number of available fences. - */ -int gem_available_fences(int fd) -{ - int num_fences; - struct drm_i915_getparam gp; - - memset(&gp, 0, sizeof(gp)); - gp.param = I915_PARAM_NUM_FENCES_AVAIL; - gp.value = &num_fences; - - num_fences = 0; - ioctl(fd, DRM_IOCTL_I915_GETPARAM, &gp, sizeof(gp)); - errno = 0; - - return num_fences; -} - bool gem_has_llc(int fd) { int has_llc; @@ -929,97 +904,6 @@ uint64_t gem_total_stolen_size(int fd) return aperture.stolen_total_size; } -/** - * gem_available_aperture_size: - * @fd: open i915 drm file descriptor - * - * Feature test macro to query the kernel for the available gpu aperture size - * usable in a batchbuffer. - * - * Returns: The available gtt address space size. - */ -uint64_t gem_available_aperture_size(int fd) -{ - struct drm_i915_gem_get_aperture aperture; - - memset(&aperture, 0, sizeof(aperture)); - aperture.aper_size = 256*1024*1024; - do_ioctl(fd, DRM_IOCTL_I915_GEM_GET_APERTURE, &aperture); - - return aperture.aper_available_size; -} - -/** - * gem_aperture_size: - * @fd: open i915 drm file descriptor - * - * Feature test macro to query the kernel for the total gpu aperture size. - * - * Returns: The total gtt address space size. - */ -uint64_t gem_aperture_size(int fd) -{ - uint64_t aperture_size = 0; - struct drm_i915_gem_context_param p; - - memset(&p, 0, sizeof(p)); - p.param = 0x3; - if (__gem_context_get_param(fd, &p) == 0) { - aperture_size = p.value; - } else { - struct drm_i915_gem_get_aperture aperture; - - memset(&aperture, 0, sizeof(aperture)); - aperture.aper_size = 256*1024*1024; - - do_ioctl(fd, DRM_IOCTL_I915_GEM_GET_APERTURE, &aperture); - aperture_size = aperture.aper_size; - } - - return aperture_size; -} - -/** - * gem_mappable_aperture_size: - * - * Feature test macro to query the kernel for the mappable gpu aperture size. - * This is the area available for GTT memory mappings. - * - * Returns: The mappable gtt address space size. - */ -uint64_t gem_mappable_aperture_size(void) -{ - struct pci_device *pci_dev = intel_get_pci_device(); - int bar; - - if (intel_gen(pci_dev->device_id) < 3) - bar = 0; - else - bar = 2; - - return pci_dev->regions[bar].size; -} - -/** - * gem_global_aperture_size: - * @fd: open i915 drm file descriptor - * - * Feature test macro to query the kernel for the global gpu aperture size. - * This is the area available for the kernel to perform address translations. - * - * Returns: The gtt address space size. - */ -uint64_t gem_global_aperture_size(int fd) -{ - struct drm_i915_gem_get_aperture aperture; - - memset(&aperture, 0, sizeof(aperture)); - aperture.aper_size = 256*1024*1024; - do_ioctl(fd, DRM_IOCTL_I915_GEM_GET_APERTURE, &aperture); - - return aperture.aper_size; -} - /** * gem_has_softpin: * @fd: open i915 drm file descriptor diff --git a/lib/ioctl_wrappers.h b/lib/ioctl_wrappers.h index b76bea564..07879ae96 100644 --- a/lib/ioctl_wrappers.h +++ b/lib/ioctl_wrappers.h @@ -117,12 +117,7 @@ bool gem_uses_full_ppgtt(int fd); int gem_gpu_reset_type(int fd); bool gem_gpu_reset_enabled(int fd); bool gem_engine_reset_enabled(int fd); -int gem_available_fences(int fd); uint64_t gem_total_stolen_size(int fd); -uint64_t gem_available_aperture_size(int fd); -uint64_t gem_aperture_size(int fd); -uint64_t gem_global_aperture_size(int fd); -uint64_t gem_mappable_aperture_size(void); bool gem_has_softpin(int fd); bool gem_has_exec_fence(int fd); diff --git a/tests/i915/gem_concurrent_all.c b/tests/i915/gem_concurrent_all.c index 9ea94125e..da850a104 100644 --- a/tests/i915/gem_concurrent_all.c +++ b/tests/i915/gem_concurrent_all.c @@ -1828,7 +1828,7 @@ igt_main c->name, s->name, "small"); igt_subtest_group { igt_fixture { - count = num_buffers(gem_mappable_aperture_size()/4, + count = num_buffers(gem_mappable_aperture_size(fd)/4, s, c, CHECK_RAM); } run_modes(name, c, modes, s, count); @@ -1839,7 +1839,7 @@ igt_main c->name, s->name, "thrash"); igt_subtest_group { igt_fixture { - count = num_buffers(gem_mappable_aperture_size(), + count = num_buffers(gem_mappable_aperture_size(fd), s, c, CHECK_RAM); } run_modes(name, c, modes, s, count); @@ -1871,7 +1871,7 @@ igt_main c->name, s->name, "shrink"); igt_subtest_group { igt_fixture { - count = num_buffers(gem_mappable_aperture_size(), + count = num_buffers(gem_mappable_aperture_size(fd), s, c, CHECK_RAM); igt_fork_shrink_helper(fd); @@ -1887,8 +1887,8 @@ igt_main c->name, s->name, "swap"); igt_subtest_group { igt_fixture { - if (intel_get_avail_ram_mb() > gem_mappable_aperture_size()/(1024*1024)) { - pin_sz = intel_get_avail_ram_mb() - gem_mappable_aperture_size()/(1024*1024); + if (intel_get_avail_ram_mb() > gem_mappable_aperture_size(fd)/(1024*1024)) { + pin_sz = intel_get_avail_ram_mb() - gem_mappable_aperture_size(fd)/(1024*1024); igt_debug("Pinning %lld MiB\n", (long long)pin_sz); pin_sz *= 1024 * 1024; @@ -1902,7 +1902,7 @@ igt_main igt_require(pinned); } - count = num_buffers(gem_mappable_aperture_size(), + count = num_buffers(gem_mappable_aperture_size(fd), s, c, CHECK_RAM | CHECK_SWAP); } run_modes(name, c, modes, s, count); diff --git a/tests/i915/gem_mmap.c b/tests/i915/gem_mmap.c index 8bad9b14e..60a64c134 100644 --- a/tests/i915/gem_mmap.c +++ b/tests/i915/gem_mmap.c @@ -53,10 +53,10 @@ test_huge_bo(int huge) switch (huge) { case -1: - huge_object_size = gem_mappable_aperture_size() / 2; + huge_object_size = gem_mappable_aperture_size(fd) / 2; break; case 0: - huge_object_size = gem_mappable_aperture_size() + PAGE_SIZE; + huge_object_size = gem_mappable_aperture_size(fd) + PAGE_SIZE; break; case 1: huge_object_size = gem_aperture_size(fd) + PAGE_SIZE; diff --git a/tests/i915/gem_mmap_gtt.c b/tests/i915/gem_mmap_gtt.c index 61fbc5bc7..528a7c726 100644 --- a/tests/i915/gem_mmap_gtt.c +++ b/tests/i915/gem_mmap_gtt.c @@ -905,7 +905,7 @@ test_huge_bo(int fd, int huge, int tiling) switch (huge) { case -1: - size = gem_mappable_aperture_size() / 2; + size = gem_mappable_aperture_size(fd) / 2; /* Power of two fence size, natural fence * alignment, and the guard page at the end @@ -920,7 +920,7 @@ test_huge_bo(int fd, int huge, int tiling) size /= 2; break; case 0: - size = gem_mappable_aperture_size() + PAGE_SIZE; + size = gem_mappable_aperture_size(fd) + PAGE_SIZE; break; default: size = gem_global_aperture_size(fd) + PAGE_SIZE; @@ -1001,13 +1001,13 @@ test_huge_copy(int fd, int huge, int tiling_a, int tiling_b, int ncpus) switch (huge) { case -2: - huge_object_size = gem_mappable_aperture_size() / 4; + huge_object_size = gem_mappable_aperture_size(fd) / 4; break; case -1: - huge_object_size = gem_mappable_aperture_size() / 2; + huge_object_size = gem_mappable_aperture_size(fd) / 2; break; case 0: - huge_object_size = gem_mappable_aperture_size() + PAGE_SIZE; + huge_object_size = gem_mappable_aperture_size(fd) + PAGE_SIZE; break; case 1: huge_object_size = gem_global_aperture_size(fd) + PAGE_SIZE; diff --git a/tests/i915/gem_pwrite.c b/tests/i915/gem_pwrite.c index d2dcc95e8..f76d2bc70 100644 --- a/tests/i915/gem_pwrite.c +++ b/tests/i915/gem_pwrite.c @@ -130,7 +130,7 @@ static void test_big_cpu(int fd, int scale, unsigned flags) switch (scale) { case 0: - size = gem_mappable_aperture_size() + 4096; + size = gem_mappable_aperture_size(fd) + 4096; break; case 1: size = gem_global_aperture_size(fd) + 4096; @@ -192,7 +192,7 @@ static void test_big_gtt(int fd, int scale, unsigned flags) igt_require(gem_mmap__has_wc(fd)); switch (scale) { case 0: - size = gem_mappable_aperture_size() + 4096; + size = gem_mappable_aperture_size(fd) + 4096; break; case 1: size = gem_global_aperture_size(fd) + 4096; @@ -257,7 +257,7 @@ static void test_random(int fd) gem_require_mmap_wc(fd); size = min(intel_get_total_ram_mb() / 2, - gem_mappable_aperture_size() + 4096); + gem_mappable_aperture_size(fd) + 4096); intel_require_memory(1, size, CHECK_RAM); handle = gem_create(fd, size); diff --git a/tests/i915/gem_shrink.c b/tests/i915/gem_shrink.c index dba62c8fa..023db8c56 100644 --- a/tests/i915/gem_shrink.c +++ b/tests/i915/gem_shrink.c @@ -439,7 +439,7 @@ igt_main * we expect the shrinker to start purging objects, * and possibly fail. */ - alloc_size = gem_mappable_aperture_size() / 2; + alloc_size = gem_mappable_aperture_size(fd) / 2; num_processes = 1 + (mem_size / (alloc_size >> 20)); igt_info("Using %d processes and %'lluMiB per process\n", diff --git a/tests/i915/gem_tiled_fence_blits.c b/tests/i915/gem_tiled_fence_blits.c index 0a633d91b..28beea898 100644 --- a/tests/i915/gem_tiled_fence_blits.c +++ b/tests/i915/gem_tiled_fence_blits.c @@ -225,7 +225,7 @@ igt_main gem_require_blitter(fd); gem_require_mappable_ggtt(fd); - count = gem_mappable_aperture_size(); /* thrash fences! */ + count = gem_mappable_aperture_size(fd); /* thrash fences! */ if (count >> 32) count = MAX_32b; count = 3 + count / (1024 * 1024); diff --git a/tests/i915/i915_pm_rpm.c b/tests/i915/i915_pm_rpm.c index 6d46c320c..6321dd403 100644 --- a/tests/i915/i915_pm_rpm.c +++ b/tests/i915/i915_pm_rpm.c @@ -1419,7 +1419,7 @@ static void gem_evict_pwrite_subtest(void) unsigned int num_trash_bos, n; uint32_t buf; - num_trash_bos = gem_mappable_aperture_size() / (1024*1024) + 1; + num_trash_bos = gem_mappable_aperture_size(drm_fd) / (1024*1024) + 1; trash_bos = malloc(num_trash_bos * sizeof(*trash_bos)); igt_assert(trash_bos); @@ -1463,7 +1463,7 @@ static bool device_in_pci_d3(void) uint16_t val; int rc; - rc = pci_device_cfg_read_u16(intel_get_pci_device(), &val, 0xd4); + rc = pci_device_cfg_read_u16(igt_device_get_pci_device(drm_fd), &val, 0xd4); igt_assert_eq(rc, 0); igt_debug("%s: PCI D3 state=%d\n", __func__, val & 0x3); diff --git a/tests/kms_big_fb.c b/tests/kms_big_fb.c index 02e9915ba..8794ace08 100644 --- a/tests/kms_big_fb.c +++ b/tests/kms_big_fb.c @@ -645,7 +645,7 @@ igt_main data.ram_size = intel_get_total_ram_mb() << 20; data.aper_size = gem_aperture_size(data.drm_fd); - data.mappable_size = gem_mappable_aperture_size(); + data.mappable_size = gem_mappable_aperture_size(data.drm_fd); igt_info("RAM: %"PRIu64" MiB, GPU address space: %"PRId64" MiB, GGTT mappable size: %"PRId64" MiB\n", data.ram_size >> 20, data.aper_size >> 20, diff --git a/tests/kms_flip.c b/tests/kms_flip.c index 51b9ac950..0f0565cf6 100755 --- a/tests/kms_flip.c +++ b/tests/kms_flip.c @@ -1282,7 +1282,7 @@ static void __run_test_on_crtc_set(struct test_output *o, int *crtc_idxs, /* 256 MB is usually the maximum mappable aperture, * (make it 4x times that to ensure failure) */ if (o->flags & TEST_BO_TOOBIG) { - bo_size = 4*gem_mappable_aperture_size(); + bo_size = 4*gem_mappable_aperture_size(drm_fd); igt_require(bo_size < gem_global_aperture_size(drm_fd)); } diff --git a/tests/prime_mmap.c b/tests/prime_mmap.c index 143342410..7c43ced85 100644 --- a/tests/prime_mmap.c +++ b/tests/prime_mmap.c @@ -447,8 +447,8 @@ test_aperture_limit(void) char *ptr1, *ptr2; uint32_t handle1, handle2; /* Two buffers the sum of which > mappable aperture */ - uint64_t size1 = (gem_mappable_aperture_size() * 7) / 8; - uint64_t size2 = (gem_mappable_aperture_size() * 3) / 8; + uint64_t size1 = (gem_mappable_aperture_size(fd) * 7) / 8; + uint64_t size2 = (gem_mappable_aperture_size(fd) * 3) / 8; handle1 = gem_create(fd, size1); fill_bo(handle1, BO_SIZE); -- 2.29.2 </div> </div> </div> </body> </html>