1 files changed, 154 insertions, 0 deletions

diff --git a/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/3162-drm-amdkfd-Save-restore-flat_scratch_lo-hi-on-gfx10.patch b/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/3162-drm-amdkfd-Save-restore-flat_scratch_lo-hi-on-gfx10.patch
new file mode 100644
index 00000000..e274e73c
--- /dev/null
+++ b/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/3162-drm-amdkfd-Save-restore-flat_scratch_lo-hi-on-gfx10.patch
@@ -0,0 +1,154 @@
+From 12596bce220b105b2ad0708ce1fe0b240fa55f5e Mon Sep 17 00:00:00 2001
+From: Jay Cornwall <jay.cornwall@amd.com>
+Date: Sun, 28 Jul 2019 15:25:05 -0500
+Subject: [PATCH 3162/4256] drm/amdkfd: Save/restore flat_scratch_lo/hi on
+ gfx10
+
+These moved from SGPRs in gfx9 to HWREG in gfx10.
+
+Cc: Shaoyun Liu <shaoyun.liu@amd.com>
+Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
+Acked-by: Alex Deucher <alexander.deucher@amd.com>
+Reviewed-by: shaoyunl <shaoyun.liu@amd.com>
+---
+ .../gpu/drm/amd/amdkfd/cwsr_trap_handler.h    | 56 +++++++++++--------
+ .../amd/amdkfd/cwsr_trap_handler_gfx10.asm    | 14 +++++
+ 2 files changed, 48 insertions(+), 22 deletions(-)
+
+diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h
+index c10e424dd1f5..8089bb37f393 100644
+--- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h
++++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h
+@@ -680,7 +680,7 @@ static const uint32_t cwsr_trap_gfx9_hex[] = {
+ };
+ 
+ static const uint32_t cwsr_trap_gfx10_hex[] = {
+-	0xbf820001, 0xbf8201b2,
++	0xbf820001, 0xbf8201c0,
+ 	0xb0804004, 0xb978f802,
+ 	0x8a788678, 0xb971f803,
+ 	0x876eff71, 0x00000400,
+@@ -772,6 +772,13 @@ static const uint32_t cwsr_trap_gfx10_hex[] = {
+ 	0xb97bf801, 0xbefe037c,
+ 	0xbefc037a, 0xf4611efa,
+ 	0xf8000000, 0x807a847a,
++	0xbefc037e, 0xb97bf814,
++	0xbefe037c, 0xbefc037a,
++	0xf4611efa, 0xf8000000,
++	0x807a847a, 0xbefc037e,
++	0xb97bf815, 0xbefe037c,
++	0xbefc037a, 0xf4611efa,
++	0xf8000000, 0x807a847a,
+ 	0xbefc037e, 0x8776ff7f,
+ 	0x04000000, 0xbeef0380,
+ 	0x886f6f76, 0xb97a2a05,
+@@ -897,7 +904,7 @@ static const uint32_t cwsr_trap_gfx10_hex[] = {
+ 	0xe0704000, 0x7a5d0000,
+ 	0x807c817c, 0x807aff7a,
+ 	0x00000080, 0xbf0a717c,
+-	0xbf85fff8, 0xbf820138,
++	0xbf85fff8, 0xbf820141,
+ 	0xbef4037e, 0x8775ff7f,
+ 	0x0000ffff, 0x8875ff75,
+ 	0x00040000, 0xbef60380,
+@@ -1033,30 +1040,35 @@ static const uint32_t cwsr_trap_gfx10_hex[] = {
+ 	0x80788478, 0xf4211e7a,
+ 	0xf0000000, 0x80788478,
+ 	0xf4211cfa, 0xf0000000,
++	0x80788478, 0xf4211bba,
++	0xf0000000, 0x80788478,
++	0xbf8cc07f, 0xb9eef814,
++	0xf4211bba, 0xf0000000,
+ 	0x80788478, 0xbf8cc07f,
+-	0xbef2036d, 0x876dff72,
+-	0x0000ffff, 0xbefc036f,
+-	0xbefe037a, 0xbeff037b,
+-	0x876f71ff, 0x000003ff,
+-	0xb9ef4803, 0xb9f9f816,
+-	0x876f71ff, 0xfffff800,
+-	0x906f8b6f, 0xb9efa2c3,
+-	0xb9f3f801, 0x876fff72,
+-	0xfc000000, 0x906f9a6f,
+-	0x8f6f906f, 0xbef30380,
++	0xb9eef815, 0xbef2036d,
++	0x876dff72, 0x0000ffff,
++	0xbefc036f, 0xbefe037a,
++	0xbeff037b, 0x876f71ff,
++	0x000003ff, 0xb9ef4803,
++	0xb9f9f816, 0x876f71ff,
++	0xfffff800, 0x906f8b6f,
++	0xb9efa2c3, 0xb9f3f801,
++	0x876fff72, 0xfc000000,
++	0x906f9a6f, 0x8f6f906f,
++	0xbef30380, 0x88736f73,
++	0x876fff72, 0x02000000,
++	0x906f996f, 0x8f6f8f6f,
+ 	0x88736f73, 0x876fff72,
+-	0x02000000, 0x906f996f,
+-	0x8f6f8f6f, 0x88736f73,
+-	0x876fff72, 0x01000000,
+-	0x906f986f, 0x8f6f996f,
+-	0x88736f73, 0x876fff70,
+-	0x00800000, 0x906f976f,
+-	0xb9f3f807, 0x87fe7e7e,
+-	0x87ea6a6a, 0xb9f0f802,
+-	0xbf8a0000, 0xbe80226c,
+-	0xbf810000, 0xbf9f0000,
++	0x01000000, 0x906f986f,
++	0x8f6f996f, 0x88736f73,
++	0x876fff70, 0x00800000,
++	0x906f976f, 0xb9f3f807,
++	0x87fe7e7e, 0x87ea6a6a,
++	0xb9f0f802, 0xbf8a0000,
++	0xbe80226c, 0xbf810000,
+ 	0xbf9f0000, 0xbf9f0000,
+ 	0xbf9f0000, 0xbf9f0000,
++	0xbf9f0000, 0x00000000,
+ };
+ static const uint32_t cwsr_trap_arcturus_hex[] = {
+ 	0xbf820001, 0xbf8202c4,
+diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
+index be6f7d1847fa..fafdfd2ac610 100644
+--- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
++++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
+@@ -132,6 +132,7 @@ var s_restore_tmp				= ttmp6
+ var s_restore_mem_offset_save			= s_restore_tmp
+ var s_restore_m0				= s_restore_alloc_size
+ var s_restore_mode				= ttmp7
++var s_restore_flat_scratch			= ttmp2
+ var s_restore_pc_lo				= ttmp0
+ var s_restore_pc_hi				= ttmp1
+ var s_restore_exec_lo				= ttmp14
+@@ -313,6 +314,12 @@ L_SAVE_HWREG:
+ 	s_getreg_b32	s_save_m0, hwreg(HW_REG_MODE)
+ 	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+ 
++	s_getreg_b32	s_save_m0, hwreg(HW_REG_SHADER_FLAT_SCRATCH_LO)
++	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
++
++	s_getreg_b32	s_save_m0, hwreg(HW_REG_SHADER_FLAT_SCRATCH_HI)
++	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
++
+ 	/* the first wave in the threadgroup */
+ 	s_and_b32	s_save_tmp, s_save_spi_init_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK
+ 	s_mov_b32	s_save_exec_hi, 0x0
+@@ -824,9 +831,16 @@ L_RESTORE_HWREG:
+ 	read_hwreg_from_mem(s_restore_trapsts, s_restore_buf_rsrc0, s_restore_mem_offset)
+ 	read_hwreg_from_mem(s_restore_xnack_mask, s_restore_buf_rsrc0, s_restore_mem_offset)
+ 	read_hwreg_from_mem(s_restore_mode, s_restore_buf_rsrc0, s_restore_mem_offset)
++	read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
++	s_waitcnt	lgkmcnt(0)
++
++	s_setreg_b32	hwreg(HW_REG_SHADER_FLAT_SCRATCH_LO), s_restore_flat_scratch
+ 
++	read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
+ 	s_waitcnt	lgkmcnt(0)						//from now on, it is safe to restore STATUS and IB_STS
+ 
++	s_setreg_b32	hwreg(HW_REG_SHADER_FLAT_SCRATCH_HI), s_restore_flat_scratch
++
+ 	s_mov_b32	s_restore_tmp, s_restore_pc_hi
+ 	s_and_b32	s_restore_pc_hi, s_restore_tmp, 0x0000ffff		//pc[47:32] //Do it here in order not to affect STATUS
+ 
+-- 
+2.17.1
+

path: root/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/3162-drm-amdkfd-Save-restore-flat_scratch_lo-hi-on-gfx10.patch

// SPDX-License-Identifier: GPL-2.0
#include <test_progs.h>
#include <sys/mman.h>
#include "test_mmap.skel.h"

struct map_data {
	__u64 val[512 * 4];
};

static size_t roundup_page(size_t sz)
{
	long page_size = sysconf(_SC_PAGE_SIZE);
	return (sz + page_size - 1) / page_size * page_size;
}

void test_mmap(void)
{
	const size_t bss_sz = roundup_page(sizeof(struct test_mmap__bss));
	const size_t map_sz = roundup_page(sizeof(struct map_data));
	const int zero = 0, one = 1, two = 2, far = 1500;
	const long page_size = sysconf(_SC_PAGE_SIZE);
	int err, duration = 0, i, data_map_fd, rdmap_fd;
	struct bpf_map *data_map, *bss_map;
	void *bss_mmaped = NULL, *map_mmaped = NULL, *tmp1, *tmp2;
	struct test_mmap__bss *bss_data;
	struct map_data *map_data;
	struct test_mmap *skel;
	__u64 val = 0;


	skel = test_mmap__open_and_load();
	if (CHECK(!skel, "skel_open_and_load", "skeleton open/load failed\n"))
		return;

	bss_map = skel->maps.bss;
	data_map = skel->maps.data_map;
	data_map_fd = bpf_map__fd(data_map);

	rdmap_fd = bpf_map__fd(skel->maps.rdonly_map);
	tmp1 = mmap(NULL, 4096, PROT_READ | PROT_WRITE, MAP_SHARED, rdmap_fd, 0);
	if (CHECK(tmp1 != MAP_FAILED, "rdonly_write_mmap", "unexpected success\n")) {
		munmap(tmp1, 4096);
		goto cleanup;
	}
	/* now double-check if it's mmap()'able at all */
	tmp1 = mmap(NULL, 4096, PROT_READ, MAP_SHARED, rdmap_fd, 0);
	if (CHECK(tmp1 == MAP_FAILED, "rdonly_read_mmap", "failed: %d\n", errno))
		goto cleanup;

	bss_mmaped = mmap(NULL, bss_sz, PROT_READ | PROT_WRITE, MAP_SHARED,
			  bpf_map__fd(bss_map), 0);
	if (CHECK(bss_mmaped == MAP_FAILED, "bss_mmap",
		  ".bss mmap failed: %d\n", errno)) {
		bss_mmaped = NULL;
		goto cleanup;
	}
	/* map as R/W first */
	map_mmaped = mmap(NULL, map_sz, PROT_READ | PROT_WRITE, MAP_SHARED,
			  data_map_fd, 0);
	if (CHECK(map_mmaped == MAP_FAILED, "data_mmap",
		  "data_map mmap failed: %d\n", errno)) {
		map_mmaped = NULL;
		goto cleanup;
	}

	bss_data = bss_mmaped;
	map_data = map_mmaped;

	CHECK_FAIL(bss_data->in_val);
	CHECK_FAIL(bss_data->out_val);
	CHECK_FAIL(skel->bss->in_val);
	CHECK_FAIL(skel->bss->out_val);
	CHECK_FAIL(map_data->val[0]);
	CHECK_FAIL(map_data->val[1]);
	CHECK_FAIL(map_data->val[2]);
	CHECK_FAIL(map_data->val[far]);

	err = test_mmap__attach(skel);
	if (CHECK(err, "attach_raw_tp", "err %d\n", err))
		goto cleanup;

	bss_data->in_val = 123;
	val = 111;
	CHECK_FAIL(bpf_map_update_elem(data_map_fd, &zero, &val, 0));

	usleep(1);

	CHECK_FAIL(bss_data->in_val != 123);
	CHECK_FAIL(bss_data->out_val != 123);
	CHECK_FAIL(skel->bss->in_val != 123);
	CHECK_FAIL(skel->bss->out_val != 123);
	CHECK_FAIL(map_data->val[0] != 111);
	CHECK_FAIL(map_data->val[1] != 222);
	CHECK_FAIL(map_data->val[2] != 123);
	CHECK_FAIL(map_data->val[far] != 3 * 123);

	CHECK_FAIL(bpf_map_lookup_elem(data_map_fd, &zero, &val));
	CHECK_FAIL(val != 111);
	CHECK_FAIL(bpf_map_lookup_elem(data_map_fd, &one, &val));
	CHECK_FAIL(val != 222);
	CHECK_FAIL(bpf_map_lookup_elem(data_map_fd, &two, &val));
	CHECK_FAIL(val != 123);
	CHECK_FAIL(bpf_map_lookup_elem(data_map_fd, &far, &val));
	CHECK_FAIL(val != 3 * 123);

	/* data_map freeze should fail due to R/W mmap() */
	err = bpf_map_freeze(data_map_fd);
	if (CHECK(!err || errno != EBUSY, "no_freeze",
		  "data_map freeze succeeded: err=%d, errno=%d\n", err, errno))
		goto cleanup;

	/* unmap R/W mapping */
	err = munmap(map_mmaped, map_sz);
	map_mmaped = NULL;
	if (CHECK(err, "data_map_munmap", "data_map munmap failed: %d\n", errno))
		goto cleanup;

	/* re-map as R/O now */
	map_mmaped = mmap(NULL, map_sz, PROT_READ, MAP_SHARED, data_map_fd, 0);
	if (CHECK(map_mmaped == MAP_FAILED, "data_mmap",
		  "data_map R/O mmap failed: %d\n", errno)) {
		map_mmaped = NULL;
		goto cleanup;
	}
	map_data = map_mmaped;

	/* map/unmap in a loop to test ref counting */
	for (i = 0; i < 10; i++) {
		int flags = i % 2 ? PROT_READ : PROT_WRITE;
		void *p;

		p = mmap(NULL, map_sz, flags, MAP_SHARED, data_map_fd, 0);
		if (CHECK_FAIL(p == MAP_FAILED))
			goto cleanup;
		err = munmap(p, map_sz);
		if (CHECK_FAIL(err))
			goto cleanup;
	}

	/* data_map freeze should now succeed due to no R/W mapping */
	err = bpf_map_freeze(data_map_fd);
	if (CHECK(err, "freeze", "data_map freeze failed: err=%d, errno=%d\n",
		  err, errno))
		goto cleanup;

	/* mapping as R/W now should fail */
	tmp1 = mmap(NULL, map_sz, PROT_READ | PROT_WRITE, MAP_SHARED,
		    data_map_fd, 0);
	if (CHECK(tmp1 != MAP_FAILED, "data_mmap", "mmap succeeded\n")) {
		munmap(tmp1, map_sz);
		goto cleanup;
	}

	bss_data->in_val = 321;
	usleep(1);
	CHECK_FAIL(bss_data->in_val != 321);
	CHECK_FAIL(bss_data->out_val != 321);
	CHECK_FAIL(skel->bss->in_val != 321);
	CHECK_FAIL(skel->bss->out_val != 321);
	CHECK_FAIL(map_data->val[0] != 111);
	CHECK_FAIL(map_data->val[1] != 222);
	CHECK_FAIL(map_data->val[2] != 321);
	CHECK_FAIL(map_data->val[far] != 3 * 321);

	/* check some more advanced mmap() manipulations */

	/* map all but last page: pages 1-3 mapped */
	tmp1 = mmap(NULL, 3 * page_size, PROT_READ, MAP_SHARED,
			  data_map_fd, 0);
	if (CHECK(tmp1 == MAP_FAILED, "adv_mmap1", "errno %d\n", errno))
		goto cleanup;

	/* unmap second page: pages 1, 3 mapped */
	err = munmap(tmp1 + page_size, page_size);
	if (CHECK(err, "adv_mmap2", "errno %d\n", errno)) {
		munmap(tmp1, map_sz);
		goto cleanup;
	}

	/* map page 2 back */
	tmp2 = mmap(tmp1 + page_size, page_size, PROT_READ,
		    MAP_SHARED | MAP_FIXED, data_map_fd, 0);
	if (CHECK(tmp2 == MAP_FAILED, "adv_mmap3", "errno %d\n", errno)) {
		munmap(tmp1, page_size);
		munmap(tmp1 + 2*page_size, page_size);
		goto cleanup;
	}
	CHECK(tmp1 + page_size != tmp2, "adv_mmap4",
	      "tmp1: %p, tmp2: %p\n", tmp1, tmp2);

	/* re-map all 4 pages */
	tmp2 = mmap(tmp1, 4 * page_size, PROT_READ, MAP_SHARED | MAP_FIXED,
		    data_map_fd, 0);
	if (CHECK(tmp2 == MAP_FAILED, "adv_mmap5", "errno %d\n", errno)) {
		munmap(tmp1, 3 * page_size); /* unmap page 1 */
		goto cleanup;
	}
	CHECK(tmp1 != tmp2, "adv_mmap6", "tmp1: %p, tmp2: %p\n", tmp1, tmp2);

	map_data = tmp2;
	CHECK_FAIL(bss_data->in_val != 321);
	CHECK_FAIL(bss_data->out_val != 321);
	CHECK_FAIL(skel->bss->in_val != 321);
	CHECK_FAIL(skel->bss->out_val != 321);
	CHECK_FAIL(map_data->val[0] != 111);
	CHECK_FAIL(map_data->val[1] != 222);
	CHECK_FAIL(map_data->val[2] != 321);
	CHECK_FAIL(map_data->val[far] != 3 * 321);

	munmap(tmp2, 4 * page_size);

	/* map all 4 pages, but with pg_off=1 page, should fail */
	tmp1 = mmap(NULL, 4 * page_size, PROT_READ, MAP_SHARED | MAP_FIXED,
		    data_map_fd, page_size /* initial page shift */);
	if (CHECK(tmp1 != MAP_FAILED, "adv_mmap7", "unexpected success")) {
		munmap(tmp1, 4 * page_size);
		goto cleanup;
	}

cleanup:
	if (bss_mmaped)
		CHECK_FAIL(munmap(bss_mmaped, bss_sz));
	if (map_mmaped)
		CHECK_FAIL(munmap(map_mmaped, map_sz));
	test_mmap__destroy(skel);
}