Skip to contentSkip to conversation

intel-decode regression testing

PATCH In review Eric Anholt 2 participants 7 messages Jan 3, 2012 – Jan 4, 2012

lore
Share & tools
lore
#0 lore raw

I tried to make this as easy as possible for updating the decode output -- it should be a matter of looking at the diff produced by make check, approving of it, and copying the -new.txt over the -ref.txt.

Tested with make check and make distcheck, including with corrupting a ref file.

#1 lore raw

PATCH 1/3 intel: Add an interface for setting the output file for decode.

Consumers often want to choose stdout vs stderr, and for testing I want to output to an open_memstream file.

---
 intel/intel_bufmgr.h |    2 ++
 intel/intel_decode.c |   14 ++++++++++++--
 2 files changed, 14 insertions(+), 2 deletions(-)
intel/intel_bufmgr.h +2 -0
index 68017a5..85da8b9 100644
@@ -34,6 +34,7 @@
 #ifndef INTEL_BUFMGR_H
 #define INTEL_BUFMGR_H
 
+#include <stdio.h>
 #include <stdint.h>
 
 struct drm_clip_rect;
@@ -199,6 +200,7 @@ void drm_intel_decode_set_dump_past_end(struct drm_intel_decode *ctx,
 					int dump_past_end);
 void drm_intel_decode_set_head_tail(struct drm_intel_decode *ctx,
 				    uint32_t head, uint32_t tail);
+void drm_intel_decode_set_output_file(struct drm_intel_decode *ctx, FILE *out);
 void drm_intel_decode(struct drm_intel_decode *ctx);
 
 
intel/intel_decode.c +12 -2
index e80e840..81ef712 100644
@@ -33,6 +33,9 @@
 
 /* Struct for tracking drm_intel_decode state. */
 struct drm_intel_decode {
+	/** stdio file where the output should land.  Defaults to stdout. */
+	FILE *out;
+
 	/** PCI device ID. */
 	uint32_t devid;
 
@@ -3558,6 +3561,7 @@ drm_intel_decode_context_alloc(uint32_t devid)
 		return NULL;
 
 	ctx->devid = devid;
+	ctx->out = stdout;
 
 	return ctx;
 }
@@ -3592,6 +3596,13 @@ drm_intel_decode_set_head_tail(struct drm_intel_decode *ctx,
 	ctx->tail = tail;
 }
 
+void
+drm_intel_decode_set_output_file(struct drm_intel_decode *ctx,
+				 FILE *out)
+{
+	ctx->out = out;
+}
+
 /**
  * Decodes an i830-i915 batch buffer, writing the output to stdout.
  *
@@ -3618,12 +3629,11 @@ drm_intel_decode(struct drm_intel_decode *ctx)
 	devid = ctx->devid;
 	head_offset = ctx->head;
 	tail_offset = ctx->tail;
+	out = ctx->out;
 
 	saved_s2_set = 0;
 	saved_s4_set = 1;
 
-	out = stdout;
-
 	while (index < count) {
 		switch ((data[index] & 0xe0000000) >> 29) {
 		case 0x0:
Signature · 1 line
1.7.7.3
#2 lore raw

PATCH 2/3 intel: Add a regression test program for intel_decode.c.

In reply to Eric Anholt
---
 configure.ac        |    2 +
 intel/.gitignore    |    1 +
 intel/Makefile.am   |    5 ++
 intel/test_decode.c |  190 +++++++++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 198 insertions(+), 0 deletions(-)
 create mode 100644 intel/.gitignore
 create mode 100644 intel/test_decode.c
configure.ac +2 -0
index 5f144bc..540622f 100644
@@ -114,6 +114,8 @@ AC_CHECK_FUNCS([clock_gettime], [CLOCK_LIB=],
                              [AC_MSG_ERROR([Couldn't find clock_gettime])])])
 AC_SUBST([CLOCK_LIB])
 
+AC_CHECK_FUNCS([open_memstream], [HAVE_OPEN_MEMSTREAM=yes])
+
 dnl Use lots of warning flags with with gcc and compatible compilers
 
 dnl Note: if you change the following variable, the cache is automatically
intel/.gitignore new file +1 -0
new file mode 100644
index 0000000..528b408
@@ -0,0 +1 @@
+test_decode
intel/Makefile.am +5 -0
index 801210f..2d3d8c4 100644
@@ -54,4 +54,9 @@ libdrm_intelincludedir = ${includedir}/libdrm
 libdrm_intelinclude_HEADERS = intel_bufmgr.h \
 			      intel_debug.h
 
+# This may be interesting even outside of "make check", due to the -dump option.
+noinst_PROGRAMS = test_decode
+
+test_decode_LDADD = libdrm_intel.la
+
 pkgconfig_DATA = libdrm_intel.pc
intel/test_decode.c new file +190 -0

190 lines not on this page · raw message

Signature · 6 lines
1.7.7.3

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw

PATCH 3/3 intel: Add regression tests for batch decode.

In reply to Eric Anholt

The .batch was generated using the dump-a-batch branch of

git://people.freedesktop.org/~anholt/mesa

using glxgears on gen7 hardware, using INTEL_DEVID_OVERRIDE for non-gen7 (this means that offsets in the buffers for non-gen7 are 0!). The .ref was generated by:

./test_decode tests/gen7-3d.batch -dump.

The .sh exists because you can't supply arguments to tests using the simple automake tests driver. Something reasonable could be done using automake's parallel-tests driver (in fact, a previous version of the patch did that), but I was concerned that:

1) The parallel-tests driver is documented to be unstable -- they may change interfaces on us later.
2) The parallel-tests driver hides the output of tests in .log files scattered all over the tree, which was ugly and more painful to work with.

---
 intel/Makefile.am                 |   17 +
 intel/tests/gen7-3d.batch         |  Bin 0 -> 4504 bytes
 intel/tests/gen7-3d.batch-ref.txt | 1350 +++++++++++++++++++++++++++++++++++++
 intel/tests/test-batch.sh         |   20 +
 4 files changed, 1387 insertions(+), 0 deletions(-)
 create mode 100644 intel/tests/gen7-3d.batch
 create mode 100644 intel/tests/gen7-3d.batch-ref.txt
 create mode 100755 intel/tests/test-batch.sh
intel/Makefile.am +17 -0
index 2d3d8c4..581c8c0 100644
@@ -57,6 +57,23 @@ libdrm_intelinclude_HEADERS = intel_bufmgr.h \
 # This may be interesting even outside of "make check", due to the -dump option.
 noinst_PROGRAMS = test_decode
 
+BATCHES = \
+	tests/gen4-3d.batch \
+	tests/gm45-3d.batch \
+	tests/gen5-3d.batch \
+	tests/gen6-3d.batch \
+	tests/gen7-3d.batch
+
+TESTS = \
+	$(BATCHES:.batch=.batch.sh)
+
+EXTRA_DIST = \
+	$(BATCHES) \
+	$(BATCHES:.batch=.batch.sh) \
+	$(BATCHES:.batch=.batch-ref.txt) \
+	$(BATCHES:.batch=.batch-ref.txt) \
+	tests/test-batch.sh
+
 test_decode_LDADD = libdrm_intel.la
 
 pkgconfig_DATA = libdrm_intel.pc
intel/tests/gen7-3d.batch new file binary +0 -0
new file mode 100644
index 0000000000000000000000000000000000000000..ee448650b946d9ba6e851708f83dde4cdd6eebb8
GIT binary patch
intel/tests/gen7-3d.batch-ref.txt new file +1350 -0

1,350 lines not on this page · raw message

intel/tests/test-batch.sh new file +20 -0
new file mode 100755
index 0000000..a94057f
@@ -0,0 +1,20 @@
+#!/bin/sh
+
+TEST_FILENAME=`echo "$0" | sed 's|.sh||'`
+./test_decode $TEST_FILENAME
+
+ret=$?
+
+# pretty-print a diff showing what happened, and leave the dumped
+# around for possibly moving over the ref.
+if test $ret = 1; then
+    REF_FILENAME="$TEST_FILENAME-ref.txt"
+    NEW_FILENAME="$TEST_FILENAME-new.txt"
+    ./test_decode $TEST_FILENAME -dump > $NEW_FILENAME
+    if test $? = 0; then
+	echo "Differences:"
+	diff -u $REF_FILENAME $NEW_FILENAME
+    fi
+fi
+
+exit $ret
Signature · 1 line
1.7.7.3
#4 lore raw

Re: [PATCH 3/3] intel: Add regression tests for batch decode.

Quoted 20 lines · Eric Anholt

The .batch was generated using the dump-a-batch branch of

git://people.freedesktop.org/~anholt/mesa

using glxgears on gen7 hardware, using INTEL_DEVID_OVERRIDE for non-gen7 (this means that offsets in the buffers for non-gen7 are 0!). The .ref was generated by:

./test_decode tests/gen7-3d.batch -dump.

The .sh exists because you can't supply arguments to tests using the simple automake tests driver. Something reasonable could be done using automake's parallel-tests driver (in fact, a previous version of the patch did that), but I was concerned that:

1) The parallel-tests driver is documented to be unstable -- they may change interfaces on us later.
2) The parallel-tests driver hides the output of tests in .log files scattered all over the tree, which was ugly and more painful to work with.

Quoted 4 lines
---
 intel/Makefile.am                 |   17 +
 intel/tests/gen7-3d.batch         |  Bin 0 -> 4504 bytes
 intel/tests/gen7-3d.batch-ref.txt | 1350 +++++++++++++++++++++++++++++++++++++

I'm missing the *.batch* stuff for gen4-gen6 mentioned in the Makefile.am.

Quoted 1 line

From looking throught the code I've also expected the *.batch to be a

symlink to test-batch.sh.

Otherwise this looks nice. I'm not really sold on this catching bugs, but it certainly helps in self-documenting patches to the decoder by including the relevant output changes in the diff.
-Daniel

Signature · 3 lines
Daniel Vetter
Mail: [email protected]
Mobile: +41 (0)79 365 57 48
#5 lore raw

Re: [PATCH 3/3] intel: Add regression tests for batch decode.

Quoted 26 lines · Daniel Vetter
Quoted 20 lines · Eric Anholt

The .batch was generated using the dump-a-batch branch of

git://people.freedesktop.org/~anholt/mesa

using glxgears on gen7 hardware, using INTEL_DEVID_OVERRIDE for non-gen7 (this means that offsets in the buffers for non-gen7 are 0!). The .ref was generated by:

./test_decode tests/gen7-3d.batch -dump.

The .sh exists because you can't supply arguments to tests using the simple automake tests driver. Something reasonable could be done using automake's parallel-tests driver (in fact, a previous version of the patch did that), but I was concerned that:

1) The parallel-tests driver is documented to be unstable -- they may change interfaces on us later.
2) The parallel-tests driver hides the output of tests in .log files scattered all over the tree, which was ugly and more painful to work with.

Quoted 4 lines
---
 intel/Makefile.am                 |   17 +
 intel/tests/gen7-3d.batch         |  Bin 0 -> 4504 bytes
 intel/tests/gen7-3d.batch-ref.txt | 1350 +++++++++++++++++++++++++++++++++++++
Quoted 3 lines

I'm missing the *.batch* stuff for gen4-gen6 mentioned in the Makefile.am. From looking throught the code I've also expected the *.batch to be a symlink to test-batch.sh.

Failure to git add. Fixed locally.

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Reviewed-by
#6 lore raw

Re: [PATCH 3/3] intel: Add regression tests for batch decode.

Quoted 27 lines · Eric Anholt
Quoted 26 lines · Daniel Vetter
Quoted 24 lines · Eric Anholt

The .batch was generated using the dump-a-batch branch of

git://people.freedesktop.org/~anholt/mesa

using glxgears on gen7 hardware, using INTEL_DEVID_OVERRIDE for non-gen7 (this means that offsets in the buffers for non-gen7 are 0!). The .ref was generated by:

./test_decode tests/gen7-3d.batch -dump.

The .sh exists because you can't supply arguments to tests using the simple automake tests driver. Something reasonable could be done using automake's parallel-tests driver (in fact, a previous version of the patch did that), but I was concerned that:

1) The parallel-tests driver is documented to be unstable -- they may change interfaces on us later.
2) The parallel-tests driver hides the output of tests in .log files scattered all over the tree, which was ugly and more painful to work with.

---
 intel/Makefile.am                 |   17 +
 intel/tests/gen7-3d.batch         |  Bin 0 -> 4504 bytes
 intel/tests/gen7-3d.batch-ref.txt | 1350 +++++++++++++++++++++++++++++++++++++
Quoted 5 lines
Quoted 3 lines

I'm missing the *.batch* stuff for gen4-gen6 mentioned in the Makefile.am. From looking throught the code I've also expected the *.batch to be a symlink to test-batch.sh.

Failure to git add. Fixed locally.

In that case, for the series:

Signature · 3 lines
Daniel Vetter
Mail: [email protected]
Mobile: +41 (0)79 365 57 48