[Intel-gfx] [PATCH v2] tests/gem_error_capture: Initial testcase for error state capture/dump

Mateo Lozano, Oscar oscar.mateo at intel.com
Mon Apr 14 12:01:12 CEST 2014


> > +	FILE *file;
> > +	int debug_fd;
> > +	char *line = NULL;
> > +	size_t line_size = 0;
> > +	char *dashes = NULL;
> > +	char *ring_name = NULL;
> > +	int bb_matched = 0;
> > +	uint32_t gtt_offset;
> > +	char expected_line[32];
> > +	int req_matched = 0;
> > +	int requests;
> > +	uint32_t seqno, tail;
> > +	long jiffies;
> > +	int ringbuf_matched = 0;
> > +	unsigned int offset, command, expected_addr = 0;
> > +	int i, items;
> > +	bool bb_ok = false, req_ok = false, ringbuf_ok = false;
> 
> Most of these are only of use in local scope.

ACK, will change.

> > +	debug_fd = igt_debugfs_open("i915_error_state", O_RDONLY);
> > +	igt_assert(debug_fd >= 0);
> > +	file = fdopen(debug_fd, "r");
> > +
> > +	while (getline(&line, &line_size, file) > 0) {
> > +		dashes = strstr(line, "---");
> > +		if (!dashes)
> > +			continue;
> > +
> > +		ring_name = realloc(ring_name, dashes - line);
> > +		strncpy(ring_name, line, dashes - line);
> > +		ring_name[dashes - line - 1] = '\0';
> > +
> > +		bb_matched = sscanf(dashes, "--- gtt_offset = 0x%08x\n",
> > +				&gtt_offset);
> > +		if (bb_matched == 1) {
> > +			igt_assert(strstr(ring_name, expected_ring_name));
> > +			igt_assert(gtt_offset == expected_offset);
> > +
> > +			for (i = 0; i < sizeof(batch) / 4; i++) {
> > +				igt_assert(getline(&line, &line_size, file) > 0);
> > +				snprintf(expected_line, sizeof(expected_line),
> "%08x :  %08x",
> > +						4*i, batch[i]);
> > +				igt_assert(strstr(line, expected_line));
> > +			}
> > +			bb_ok = true;
> > +			continue;
> > +		}
> > +
> > +		req_matched = sscanf(dashes, "--- %d requests\n",
> > +				&requests);
> > +		if (req_matched == 1) {
> > +			igt_assert(strstr(ring_name, expected_ring_name));
> > +			igt_assert(requests == 1);
> 
> Bad assumption. You could still have the request from
> gem_quiescent_gpu() which may not have been retired before the error
> triggers.

Hmmmm... but I make a first valid gem_execbuf()+gem_sync() before the one that actually hangs. Shouldn´t that make sure that all previous requests have been retired?

> > +
> > +			igt_assert(getline(&line, &line_size, file) > 0);
> > +			items = sscanf(line, "  seqno 0x%08x, emitted %ld, tail
> 0x%08x\n",
> > +					&seqno, &jiffies, &tail);
> > +			igt_assert(items == 3);
> 
> Bad. I may change the format. s/may/will/

I still need to make sure I got a valid tail, so I need to know the format and fail if don´t understand it. Or do you want me to use a different tail, maybe the ringbuffer one?

> > +			req_ok = true;
> > +			continue;
> > +		}
> > +
> > +		ringbuf_matched = sscanf(dashes, "--- ringbuffer = 0x%08x\n",
> > +				&gtt_offset);
> > +		if (ringbuf_matched == 1) {
> > +			if (!strstr(ring_name, expected_ring_name))
> > +				continue;
> > +			igt_assert(req_ok);
> > +
> > +			for (i = 0; i < tail / 4; i++) {
> > +				igt_assert(getline(&line, &line_size, file) > 0);
> > +				items = sscanf(line, "%08x :  %08x\n",
> > +						&offset, &command);
> > +				igt_assert(items == 2);
> > +				if ((command & 0x1F800000) ==
> MI_BATCH_BUFFER_START) {
> > +					igt_assert(getline(&line, &line_size,
> file) > 0);
> > +					items = sscanf(line, "%08x :  %08x\n",
> > +							&offset,
> &expected_addr);
> > +					igt_assert(items == 2);
> > +					i++;
> > +				}
> > +			}
> > +			igt_assert(expected_addr == expected_offset);
> 
> Bad. Some gen encode flags into the BB start address.

ACK, will change.

> > +			ringbuf_ok = true;
> > +			continue;
> > +		}
> > +
> > +		if (bb_ok && req_ok && ringbuf_ok)
> > +			break;
> > +	}
> > +	igt_assert(bb_ok && req_ok && ringbuf_ok);
> > +
> > +	free(line);
> > +	free(ring_name);
> > +	close(debug_fd);
> > +}



More information about the Intel-gfx mailing list