Thread (54 messages) 54 messages, 5 authors, 2021-06-10

Re: [igt-dev] [PATCH i-g-t 14/17] Verify execbuf ok with stale prot-buff and regular context

From: Teres Alexis, Alan Previn <hidden>
Date: 2021-06-05 00:27:08

Will do - thanks.

-----Original Message-----
From: Vivi, Rodrigo <rodrigo.vivi@intel.com> 
Sent: Friday, June 4, 2021 6:56 AM
To: Teres Alexis, Alan Previn <redacted>
Cc: igt-dev@lists.freedesktop.org
Subject: Re: [igt-dev] [PATCH i-g-t 14/17] Verify execbuf ok with stale prot-buff and regular context

On Tue, May 18, 2021 at 03:33:41AM -0700, Alan Previn wrote:
Add a check to verify that reusing a stale protected buffer in a 
gem_execbuff call, but using a regular (not-
protcted) context will succeed despite after a teardown
       ^ typo

with this fixed:
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
quoted hunk ↗ jump to hunk
(triggered by suspend-resume cycle).

This ensures that user space applications that choose not to opt-in 
for strict PXP teardown awareness (by using a regular context) won't 
suffer gem_execbuff failures if a protected buffer was among the 
assets used in any of its rendering operations.

Reuse the existing subtest to minimize the number of suspend resume 
cycles we take.

Signed-off-by: Alan Previn <redacted>
---
 tests/i915/gem_pxp.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)
diff --git a/tests/i915/gem_pxp.c b/tests/i915/gem_pxp.c index 
b4f369ae..e3c21c70 100644
--- a/tests/i915/gem_pxp.c
+++ b/tests/i915/gem_pxp.c
@@ -760,7 +760,7 @@ static void free_exec_assets(int i915, struct 
simple_exec_assets *data)  static void 
test_pxp_pwrcycle_staleasset_execution(int i915, struct powermgt_data *pm)  {
 	int ret;
-	struct simple_exec_assets data[2] = {{0}, {0}};
+	struct simple_exec_assets data[3] = {{0}, {0}, {0}};
 	uint32_t ctx2;
 	struct intel_bb *ibb2;
 
@@ -777,6 +777,15 @@ static void test_pxp_pwrcycle_staleasset_execution(int i915, struct powermgt_dat
 	ret = gem_execbuf_flush_store_dw(i915, data[1].ibb, data[1].ctx, data[1].fencebuf);
 	igt_assert(ret == 0);
 
+	/*
+	 * For data[2]: Use non-pxp contexts but with protected buffers
+	 * to verify the "pxp-execution-opt-in" expectation where
+	 * execution will run without error (not buffer)
+	 */
+	prepare_exec_assets(i915, &data[2], false, true);
+	ret = gem_execbuf_flush_store_dw(i915, data[2].ibb, data[2].ctx, data[2].fencebuf);
+	igt_assert(ret == 0);
+
 	trigger_powermgt_suspend_cycle(i915, pm);
 
 	ret = gem_execbuf_flush_store_dw(i915, data[0].ibb, data[0].ctx, 
data[0].fencebuf); @@ -797,8 +806,12 @@ static void test_pxp_pwrcycle_staleasset_execution(int i915, struct powermgt_dat
 	ret = gem_execbuf_flush_store_dw(i915, ibb2, ctx2, data[1].fencebuf);
 	igt_assert_f((ret == -ENOEXEC), "Executing stale pxp buffer didn't 
fail with -ENOEXEC\n");
 
+	ret = gem_execbuf_flush_store_dw(i915, data[2].ibb, data[2].ctx, data[2].fencebuf);
+	igt_assert_f((ret == 0), "Opt-out-execution with stale pxp buffer 
+didn't succeed\n");
+
 	free_exec_assets(i915, &data[0]);
 	free_exec_assets(i915, &data[1]);
+	free_exec_assets(i915, &data[2]);
 }
 
 igt_main
--
2.25.1

_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help