[PATCH i-g-t] igt/gem_exec_capture: Capture many, many objects

[Chris Wilson <chris@chris-wilson.co.uk>]

Exercise O(N^2) behaviour in reading the error state, and push it to the
extreme.

Reported-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 lib/meson.build              |   1 +
 tests/gem_exec_capture.c     | 320 ++++++++++++++++++++++++++++++++++-
 tests/intel-ci/blacklist.txt |   1 +
 3 files changed, 319 insertions(+), 3 deletions(-)

diff --git a/lib/meson.build b/lib/meson.build
index e60e5e02f..ce5465b63 100644
--- a/lib/meson.build
+++ b/lib/meson.build
@@ -66,6 +66,7 @@ lib_deps = [
 	math,
 	realtime,
 	ssl,
+	zlib
 ]
 
 if libdrm_intel.found()
diff --git a/tests/gem_exec_capture.c b/tests/gem_exec_capture.c
index 2dc06ce43..2d0839468 100644
--- a/tests/gem_exec_capture.c
+++ b/tests/gem_exec_capture.c
@@ -21,8 +21,11 @@
  * IN THE SOFTWARE.
  */
 
+#include <zlib.h>
+
 #include "igt.h"
 #include "igt_device.h"
+#include "igt_rand.h"
 #include "igt_sysfs.h"
 
 #define LOCAL_OBJECT_CAPTURE (1 << 7)
@@ -54,10 +57,11 @@ static void check_error_state(int dir, struct drm_i915_gem_exec_object2 *obj)
 		found = true;
 	}
 
+	free(error);
 	igt_assert(found);
 }
 
-static void __capture(int fd, int dir, unsigned ring, uint32_t target)
+static void __capture1(int fd, int dir, unsigned ring, uint32_t target)
 {
 	const int gen = intel_gen(intel_get_drm_devid(fd));
 	struct drm_i915_gem_exec_object2 obj[4];
@@ -167,10 +171,295 @@ static void capture(int fd, int dir, unsigned ring)
 	uint32_t handle;
 
 	handle = gem_create(fd, 4096);
-	__capture(fd, dir, ring, handle);
+	__capture1(fd, dir, ring, handle);
 	gem_close(fd, handle);
 }
 
+static uint64_t *__captureN(int fd, int dir, unsigned ring,
+			    unsigned int size, int count,
+			    unsigned int flags)
+#define INCREMENTAL 0x1
+{
+	const int gen = intel_gen(intel_get_drm_devid(fd));
+	struct drm_i915_gem_exec_object2 *obj;
+	struct drm_i915_gem_relocation_entry reloc[2];
+	struct drm_i915_gem_execbuffer2 execbuf;
+	uint32_t *batch, *seqno;
+	uint64_t *offsets;
+	int i;
+
+	offsets = calloc(count , sizeof(*offsets));
+	igt_assert(offsets);
+
+	obj = calloc(count + 2, sizeof(*obj));
+	igt_assert(obj);
+
+	obj[0].handle = gem_create(fd, 4096);
+	for (i = 0; i < count; i++) {
+		obj[i + 1].handle = gem_create(fd, size);
+		obj[i + 1].flags =
+			LOCAL_OBJECT_CAPTURE | EXEC_OBJECT_SUPPORTS_48B_ADDRESS;
+		if (flags & INCREMENTAL) {
+			uint32_t *ptr;
+
+			ptr = gem_mmap__cpu(fd, obj[i + 1].handle,
+					    0, size, PROT_WRITE);
+			for (unsigned int n = 0; n < size / sizeof(*ptr); n++)
+				ptr[n] = i * size + n;
+			munmap(ptr, size);
+		}
+	}
+
+	obj[count + 1].handle = gem_create(fd, 4096);
+	obj[count + 1].relocs_ptr = (uintptr_t)reloc;
+	obj[count + 1].relocation_count = ARRAY_SIZE(reloc);
+
+	memset(reloc, 0, sizeof(reloc));
+	reloc[0].target_handle = obj[count + 1].handle; /* recurse */
+	reloc[0].presumed_offset = 0;
+	reloc[0].offset = 5*sizeof(uint32_t);
+	reloc[0].delta = 0;
+	reloc[0].read_domains = I915_GEM_DOMAIN_COMMAND;
+	reloc[0].write_domain = 0;
+
+	reloc[1].target_handle = obj[0].handle; /* breadcrumb */
+	reloc[1].presumed_offset = 0;
+	reloc[1].offset = sizeof(uint32_t);
+	reloc[1].delta = 0;
+	reloc[1].read_domains = I915_GEM_DOMAIN_RENDER;
+	reloc[1].write_domain = I915_GEM_DOMAIN_RENDER;
+
+	seqno = gem_mmap__wc(fd, obj[0].handle, 0, 4096, PROT_READ);
+	gem_set_domain(fd, obj[0].handle,
+			I915_GEM_DOMAIN_GTT, I915_GEM_DOMAIN_GTT);
+
+	batch = gem_mmap__cpu(fd, obj[count + 1].handle, 0, 4096, PROT_WRITE);
+	gem_set_domain(fd, obj[count + 1].handle,
+			I915_GEM_DOMAIN_CPU, I915_GEM_DOMAIN_CPU);
+
+	i = 0;
+	batch[i] = MI_STORE_DWORD_IMM | (gen < 6 ? 1 << 22 : 0);
+	if (gen >= 8) {
+		batch[++i] = 0;
+		batch[++i] = 0;
+	} else if (gen >= 4) {
+		batch[++i] = 0;
+		batch[++i] = 0;
+		reloc[1].offset += sizeof(uint32_t);
+	} else {
+		batch[i]--;
+		batch[++i] = 0;
+	}
+	batch[++i] = 0xc0ffee;
+	if (gen < 3)
+		batch[++i] = MI_NOOP;
+
+	batch[++i] = MI_BATCH_BUFFER_START; /* not crashed? try again! */
+	if (gen >= 8) {
+		batch[i] |= 1 << 8 | 1;
+		batch[++i] = 0;
+		batch[++i] = 0;
+	} else if (gen >= 6) {
+		batch[i] |= 1 << 8;
+		batch[++i] = 0;
+	} else {
+		batch[i] |= 2 << 6;
+		batch[++i] = 0;
+		if (gen < 4) {
+			batch[i] |= 1;
+			reloc[0].delta = 1;
+		}
+	}
+	munmap(batch, 4096);
+
+	memset(&execbuf, 0, sizeof(execbuf));
+	execbuf.buffers_ptr = (uintptr_t)obj;
+	execbuf.buffer_count = count + 2;
+	execbuf.flags = ring;
+	if (gen > 3 && gen < 6)
+		execbuf.flags |= I915_EXEC_SECURE;
+	gem_execbuf(fd, &execbuf);
+
+	/* Wait for the request to start */
+	while (*(volatile uint32_t *)seqno != 0xc0ffee)
+		igt_assert(gem_bo_busy(fd, obj[0].handle));
+	munmap(seqno, 4096);
+
+	igt_force_gpu_reset(fd);
+
+	gem_sync(fd, obj[count + 1].handle);
+	gem_close(fd, obj[count + 1].handle);
+	for (i = 0; i < count; i++) {
+		offsets[i] = obj[i + 1].offset;
+		gem_close(fd, obj[i + 1].handle);
+	}
+	gem_close(fd, obj[0].handle);
+
+	return offsets;
+}
+
+static unsigned long zlib_inflate(uint32_t **ptr, unsigned long len)
+{
+	struct z_stream_s zstream;
+	void *out;
+
+	memset(&zstream, 0, sizeof(zstream));
+
+	zstream.next_in = (unsigned char *)*ptr;
+	zstream.avail_in = 4*len;
+
+	if (inflateInit(&zstream) != Z_OK)
+		return 0;
+
+	out = malloc(128*4096); /* approximate obj size */
+	zstream.next_out = out;
+	zstream.avail_out = 128*4096;
+
+	do {
+		switch (inflate(&zstream, Z_SYNC_FLUSH)) {
+		case Z_STREAM_END:
+			goto end;
+		case Z_OK:
+			break;
+		default:
+			inflateEnd(&zstream);
+			return 0;
+		}
+
+		if (zstream.avail_out)
+			break;
+
+		out = realloc(out, 2*zstream.total_out);
+		if (out == NULL) {
+			inflateEnd(&zstream);
+			return 0;
+		}
+
+		zstream.next_out = (unsigned char *)out + zstream.total_out;
+		zstream.avail_out = zstream.total_out;
+	} while (1);
+end:
+	inflateEnd(&zstream);
+	free(*ptr);
+	*ptr = out;
+	return zstream.total_out / 4;
+}
+
+static unsigned long
+ascii85_decode(const char *in, uint32_t **out, bool inflate)
+{
+	unsigned long len = 0, size = 1024;
+
+	*out = realloc(*out, sizeof(uint32_t)*size);
+	if (*out == NULL)
+		return 0;
+
+	while (*in >= '!' && *in <= 'z') {
+		uint32_t v = 0;
+
+		if (len == size) {
+			size *= 2;
+			*out = realloc(*out, sizeof(uint32_t)*size);
+			if (*out == NULL)
+				return 0;
+		}
+
+		if (*in == 'z') {
+			in++;
+		} else {
+			v += in[0] - 33; v *= 85;
+			v += in[1] - 33; v *= 85;
+			v += in[2] - 33; v *= 85;
+			v += in[3] - 33; v *= 85;
+			v += in[4] - 33;
+			in += 5;
+		}
+		(*out)[len++] = v;
+	}
+
+	if (!inflate)
+		return len;
+
+	return zlib_inflate(out, len);
+}
+
+static void many(int fd, int dir, uint64_t size, unsigned int flags)
+{
+	uint64_t ram, gtt;
+	unsigned long count, blobs;
+	char *error, *str;
+	uint64_t *offsets;
+
+	gtt = gem_aperture_size(fd) / size;
+	ram = (intel_get_avail_ram_mb() << 20) / size;
+	igt_debug("Available objects in GTT:%"PRIu64", RAM:%"PRIu64"\n",
+		  gtt, ram);
+
+	count = min(gtt, ram) / 4;
+	igt_require(count > 1);
+
+	intel_require_memory(count, size, CHECK_RAM);
+
+	offsets = __captureN(fd, dir, 0, size, count, flags);
+
+	error = igt_sysfs_get(dir, "error");
+	igt_sysfs_set(dir, "error", "Begone!");
+	igt_assert(error);
+
+	blobs = 0;
+	/* render ring --- user = 0x00000000 ffffd000 */
+	for (str = error; (str = strstr(str, "--- user = ")); str++) {
+		uint32_t *data = NULL;
+		unsigned long idx, sz;
+		uint32_t expect;
+		uint32_t hi, lo;
+		uint64_t addr;
+
+		igt_assert(sscanf(str, "--- user = 0x%x %x", &hi, &lo) == 2);
+		addr = hi;
+		addr <<= 32;
+		addr |= lo;
+		for (idx = 0; idx < count; idx++) {
+			if (offsets[idx] == addr)
+				break;
+		}
+		igt_assert(idx < count);
+		igt_debug("offset:%"PRIx64", index:%ld\n", addr, idx);
+
+		str = strchr(str, '\n');
+		if (!str)
+			break;
+
+		str++;
+		if (!(*str == ':' || *str == '~'))
+			continue;
+
+		if (!strchr(str, '\n'))
+			break; /* unterminated -> ENOMEM */
+
+		igt_debug("blob:%.64s\n", str);
+		sz = ascii85_decode(str + 1, &data, *str == ':');
+		igt_assert_eq(4 * sz, size);
+
+		expect = 0;
+		if (flags & INCREMENTAL)
+			expect = idx * size;
+		for (idx = 0; idx < sz; idx++) {
+			igt_assert_eq(data[idx], expect);
+			if (flags & INCREMENTAL)
+				expect++;
+		}
+
+		blobs++;
+		free(data);
+	}
+	igt_info("Captured %lu %"PRId64"-blobs out of a total of %lu\n",
+		 blobs, size >> 12, count);
+
+	free(error);
+	free(offsets);
+}
+
 static void userptr(int fd, int dir)
 {
 	uint32_t handle;
@@ -179,7 +468,7 @@ static void userptr(int fd, int dir)
 	igt_assert(posix_memalign(&ptr, 4096, 4096) == 0);
 	igt_require(__gem_userptr(fd, ptr, 4096, 0, 0, &handle) == 0);
 
-	__capture(fd, dir, 0, handle);
+	__capture1(fd, dir, 0, handle);
 
 	gem_close(fd, handle);
 	free(ptr);
@@ -236,6 +525,31 @@ igt_main
 		}
 	}
 
+	igt_subtest_f("many-4K-zero") {
+		igt_require(gem_can_store_dword(fd, 0));
+		many(fd, dir, 1<<12, 0);
+	}
+
+	igt_subtest_f("many-4K-incremental") {
+		igt_require(gem_can_store_dword(fd, 0));
+		many(fd, dir, 1<<12, INCREMENTAL);
+	}
+
+	igt_subtest_f("many-2M-zero") {
+		igt_require(gem_can_store_dword(fd, 0));
+		many(fd, dir, 2<<20, 0);
+	}
+
+	igt_subtest_f("many-2M-incremental") {
+		igt_require(gem_can_store_dword(fd, 0));
+		many(fd, dir, 2<<20, INCREMENTAL);
+	}
+
+	igt_subtest_f("many-256M-incremental") {
+		igt_require(gem_can_store_dword(fd, 0));
+		many(fd, dir, 256<<20, INCREMENTAL);
+	}
+
 	/* And check we can read from different types of objects */
 
 	igt_subtest_f("userptr") {
diff --git a/tests/intel-ci/blacklist.txt b/tests/intel-ci/blacklist.txt
index 88b2fe313..fa33264df 100644
--- a/tests/intel-ci/blacklist.txt
+++ b/tests/intel-ci/blacklist.txt
@@ -28,6 +28,7 @@ igt@gem_ctx_thrash(@.*)?
 igt@gem_evict_alignment(@.*)?
 igt@gem_evict_everything(@.*)?
 igt@gem_exec_alignment@(?!.*single).*
+igt@gem_exec_capture@many-(?!4K-).*
 igt@gem_exec_fence@(?!.*basic).*
 igt@gem_exec_flush@(?!.*basic).*
 igt@gem_exec_gttfill@(?!.*basic).*
-- 
2.19.0.rc2

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx