Skip to contentSkip to conversation

strace/drm: Add i915 ioctls to strace

RFC In review Patrik Jakobsson 6 participants 12 messages May 6, 2015 – May 19, 2015 2 messages not in the archive Kbuild

lore
Share & tools
lore
#0 lore raw

cover strace/drm: Add i915 ioctls to strace

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Signature · 6 lines
2.1.0

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

RFC 1/2 strace/drm: Print extended info for drm and i915 ioctls

This adds a dispatcher for extending drm ioctl debugging info and adds the i915 ioctls to the xlat framework.

---
 Makefile.am             |   2 +
 defs.h                  |   2 +
 drm.c                   | 104 ++++++++++++++++++++++++++++++++++++++++++++++++
 drm_i915.c              |  43 ++++++++++++++++++++
 ioctl.c                 |   5 +++
 xlat/drm_i915_ioctls.in |  51 ++++++++++++++++++++++++
 6 files changed, 207 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_ioctls.in
Makefile.am +2 -0
index 549aebc..941e12a 100644
@@ -121,6 +121,8 @@ strace_SOURCES =	\
 	utime.c		\
 	utimes.c	\
 	v4l2.c		\
+	drm.c		\
+	drm_i915.c	\
 	vsprintf.c	\
 	wait.c		\
 	xattr.c
defs.h +2 -0
index 77c819c..699e244 100644
@@ -571,6 +571,8 @@ extern int sock_ioctl(struct tcb *, const unsigned int, long);
 extern int term_ioctl(struct tcb *, const unsigned int, long);
 extern int ubi_ioctl(struct tcb *, const unsigned int, long);
 extern int v4l2_ioctl(struct tcb *, const unsigned int, long);
+extern int drm_ioctl(struct tcb *, const unsigned int, long);
+extern int drm_i915_ioctl(struct tcb *, const unsigned int, long);
 
 extern int tv_nz(const struct timeval *);
 extern int tv_cmp(const struct timeval *, const struct timeval *);
drm.c new file +104 -0

104 lines not on this page · raw message

drm_i915.c new file +43 -0

43 lines not on this page · raw message

ioctl.c +5 -0
index c67d048..225e850 100644
@@ -216,6 +216,8 @@ ioctl_decode_command_number(unsigned int arg)
 				return 1;
 			}
 			return 0;
+		case 'd': /* DRM */
+			return 1;
 		default:
 			return 0;
 	}
@@ -252,6 +254,8 @@ ioctl_decode(struct tcb *tcp, unsigned int code, long arg)
 		return ubi_ioctl(tcp, code, arg);
 	case 'V':
 		return v4l2_ioctl(tcp, code, arg);
+	case 'd':
+		return drm_ioctl(tcp, code, arg);
 	case '=':
 		return ptp_ioctl(tcp, code, arg);
 #ifdef HAVE_LINUX_INPUT_H
@@ -284,6 +288,7 @@ ioctl_decode(struct tcb *tcp, unsigned int code, long arg)
  *   d	sys/des.h			(possible overlap)
  *   d	vax/dkio.h			(possible overlap)
  *   d	vaxuba/rxreg.h			(possible overlap)
+ *   d  drm/drm.h
  *   f	sys/filio.h
  *   g	sunwindow/win_ioctl.h		-no overlap-
  *   g	sunwindowdev/winioctl.c		!no manifest constant! -no overlap-
xlat/drm_i915_ioctls.in new file +51 -0

51 lines not on this page · raw message

Signature · 6 lines
2.1.0

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

A message that is not in the archive (sent off-list or not yet received): 1430923683-17741-1-git-send-email-patrik.jakobsson-VuQAYsv1563Yd54FQh9/[email protected]

#3 lore raw

RFC 2/2 strace/drm: Print args for most common i915 ioctls

---
 drm_i915.c                 | 235 +++++++++++++++++++++++++++++++++++++++++++++
 xlat/drm_i915_getparams.in |  28 ++++++
 xlat/drm_i915_setparams.in |   4 +
 3 files changed, 267 insertions(+)
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_setparams.in
drm_i915.c +235 -0
index 860e22b..a93f01f 100644
@@ -33,11 +33,246 @@
 #include <drm/drm.h>
 #include <drm/i915_drm.h>
 #include "xlat/drm_i915_ioctls.h"
+#include "xlat/drm_i915_getparams.h"
+#include "xlat/drm_i915_setparams.h"
+
+static int i915_getparam(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_getparam param;
+	int value;
+
+	if (entering(tcp) || umove(tcp, arg, &param))
+		return 0;
+	if (umove(tcp, (long)param.value, &value))
+		return 0;
+
+	tprintf(", {param=");
+	printxval(drm_i915_getparams, param.param, "I915_PARAM_???");
+	tprintf(", value=");
+	switch (param.param) {
+	case I915_PARAM_CHIPSET_ID:
+		tprintf("0x%04x", value);
+		break;
+	default:
+		tprintf("%d", value);
+	}
+	tprintf("}");
+
+	return 1;
+}
+
+static int i915_setparam(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_setparam param;
+
+	if (exiting(tcp) || umove(tcp, arg, &param))
+		return 0;
+
+	tprintf(", {param=");
+	printxval(drm_i915_setparams, param.param, "I915_PARAM_???");
+	tprintf(", value=%d}", param.value);
+
+	return 1;
+}
+
+static int i915_gem_execbuffer2(struct tcb *tcp, const unsigned int code,
+				long arg)
+{
+	struct drm_i915_gem_execbuffer2 eb;
+
+	if (exiting(tcp) || umove(tcp, arg, &eb))
+		return 0;
+
+	tprintf(", {buffers_ptr=%p, buffer_count=%u, "
+		"batch_start_offset=%x, batch_len=%u, DR1=%u, DR4=%u, "
+		"num_cliprects=%u, cliprects_ptr=%p, flags=0x%Lx}",
+		(void *)eb.buffers_ptr, eb.buffer_count, eb.batch_start_offset,
+		eb.batch_len, eb.DR1, eb.DR4, eb.num_cliprects,
+		(void *)eb.cliprects_ptr, eb.flags);
+
+	return 1;
+}
+
+static int i915_gem_busy(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_busy busy;
+
+	if (entering(tcp) || umove(tcp, arg, &busy))
+		return 0;
+
+	tprintf(", {handle=%u, busy=%c, ring=%u}",
+		busy.handle, (busy.busy & 0x1) ? 'Y' : 'N', (busy.busy >> 16));
+
+	return 1;
+}
+
+static int i915_gem_create(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_create create;
+
+	if (entering(tcp) || umove(tcp, arg, &create))
+		return 0;
+
+	tprintf(", {size=%Lu, handle=%u}", create.size, create.handle);
+
+	return 1;
+}
+
+static int i915_gem_pread(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_pread pr;
+
+	if (exiting(tcp) || umove(tcp, arg, &pr))
+		return 0;
+
+	tprintf(", {handle=%u, offset=%Lu, size=%Lu, data_ptr=%p}",
+		pr.handle, pr.offset, pr.size, (void *)pr.data_ptr);
+
+	return 1;
+}
+
+static int i915_gem_pwrite(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_pwrite pw;
+
+	if (exiting(tcp) || umove(tcp, arg, &pw))
+		return 0;
+
+	tprintf(", {handle=%u, offset=%Lu, size=%Lu, data_ptr=%p}",
+		pw.handle, pw.offset, pw.size, (void *)pw.data_ptr);
+
+	return 1;
+}
+
+static int i915_gem_mmap(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_mmap mmap;
+
+	if (entering(tcp) || umove(tcp, arg, &mmap))
+		return 0;
+
+	tprintf(", {handle=%u, offset=%Lu, size=%Lu, addr_ptr=%p}",
+		mmap.handle, mmap.offset, mmap.size, (void *)mmap.addr_ptr);
+
+	return 1;
+}
+
+static int i915_gem_mmap_gtt(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_mmap_gtt mmap;
+
+	if (entering(tcp) || umove(tcp, arg, &mmap))
+		return 0;
+
+	tprintf(", {handle=%u, offset=%Lu}", mmap.handle, mmap.offset);
+
+	return 1;
+}
+
+static int i915_gem_set_domain(struct tcb *tcp, const unsigned int code,
+			       long arg)
+{
+	struct drm_i915_gem_set_domain dom;
+
+	if (entering(tcp) || umove(tcp, arg, &dom))
+		return 0;
+
+	tprintf(", {handle=%u, read_domains=%x, write_domain=%x}",
+		dom.handle, dom.read_domains, dom.write_domain);
+
+	return 1;
+}
+
+static int i915_gem_madvise(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_madvise madv;
+
+	if (entering(tcp) || umove(tcp, arg, &madv))
+		return 0;
+
+	tprintf(", {handle=%u, madv=%u, retained=%u}",
+		madv.handle, madv.madv, madv.retained);
+
+	return 1;
+}
+
+static int i915_gem_get_tiling(struct tcb *tcp, const unsigned int code,
+			       long arg)
+{
+	struct drm_i915_gem_get_tiling tiling;
+
+	if (entering(tcp) || umove(tcp, arg, &tiling))
+		return 0;
+
+	tprintf(", {handle=%u, tiling_mode=%u, swizzle_mode=%u}",
+		tiling.handle, tiling.tiling_mode, tiling.swizzle_mode);
+
+	return 1;
+}
+
+static int i915_gem_set_tiling(struct tcb *tcp, const unsigned int code,
+			       long arg)
+{
+	struct drm_i915_gem_set_tiling tiling;
+
+	if (entering(tcp) || umove(tcp, arg, &tiling))
+		return 0;
+
+	tprintf(", {handle=%u, tiling_mode=%u, stride=%u, swizzle_mode=%u}",
+		tiling.handle, tiling.tiling_mode, tiling.stride,
+		tiling.swizzle_mode);
+
+	return 1;
+}
+
+static int i915_gem_userptr(struct tcb *tcp, const unsigned int code, long arg)
+{
+	struct drm_i915_gem_userptr uptr;
+
+	if (entering(tcp) || umove(tcp, arg, &uptr))
+		return 0;
+
+	tprintf(", {user_ptr=%p, user_size=%Lu, flags=0x%x, handle=%u}",
+		(void *)uptr.user_ptr, uptr.user_size, uptr.flags, uptr.handle);
+
+	return 1;
+}
 
 int drm_i915_ioctl(struct tcb *tcp, const unsigned int code, long arg)
 {
 	if (entering(tcp))
 		printxval(drm_i915_ioctls, code, "I915_IOCTL_???");
 
+	switch (code) {
+	case DRM_IOCTL_I915_GETPARAM:
+		return i915_getparam(tcp, code, arg);
+	case DRM_IOCTL_I915_SETPARAM:
+		return i915_setparam(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_EXECBUFFER2:
+		return i915_gem_execbuffer2(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_BUSY:
+		return i915_gem_busy(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_CREATE:
+		return i915_gem_create(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_PREAD:
+		return i915_gem_pread(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_PWRITE:
+		return i915_gem_pwrite(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_MMAP:
+		return i915_gem_mmap(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_MMAP_GTT:
+		return i915_gem_mmap_gtt(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_SET_DOMAIN:
+		return i915_gem_set_domain(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_MADVISE:
+		return i915_gem_madvise(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_GET_TILING:
+		return i915_gem_get_tiling(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_SET_TILING:
+		return i915_gem_set_tiling(tcp, code, arg);
+	case DRM_IOCTL_I915_GEM_USERPTR:
+		return i915_gem_userptr(tcp, code, arg);
+	}
+
 	return 0;
 }
xlat/drm_i915_getparams.in new file +28 -0
new file mode 100644
index 0000000..15275cb
@@ -0,0 +1,28 @@
+I915_PARAM_IRQ_ACTIVE
+I915_PARAM_ALLOW_BATCHBUFFER
+I915_PARAM_LAST_DISPATCH
+I915_PARAM_CHIPSET_ID
+I915_PARAM_HAS_GEM
+I915_PARAM_NUM_FENCES_AVAIL
+I915_PARAM_HAS_OVERLAY
+I915_PARAM_HAS_PAGEFLIPPING
+I915_PARAM_HAS_EXECBUF2
+I915_PARAM_HAS_BSD
+I915_PARAM_HAS_BLT
+I915_PARAM_HAS_RELAXED_FENCING
+I915_PARAM_HAS_COHERENT_RINGS
+I915_PARAM_HAS_EXEC_CONSTANTS
+I915_PARAM_HAS_RELAXED_DELTA
+I915_PARAM_HAS_GEN7_SOL_RESET
+I915_PARAM_HAS_LLC
+I915_PARAM_HAS_ALIASING_PPGTT
+I915_PARAM_HAS_WAIT_TIMEOUT
+I915_PARAM_HAS_SEMAPHORES
+I915_PARAM_HAS_PRIME_VMAP_FLUSH
+I915_PARAM_HAS_VEBOX
+I915_PARAM_HAS_SECURE_BATCHES
+I915_PARAM_HAS_PINNED_BATCHES
+I915_PARAM_HAS_EXEC_NO_RELOC
+I915_PARAM_HAS_EXEC_HANDLE_LUT
+I915_PARAM_HAS_WT
+I915_PARAM_CMD_PARSER_VERSION
xlat/drm_i915_setparams.in new file +4 -0
new file mode 100644
index 0000000..d93d2ea
@@ -0,0 +1,4 @@
+I915_SETPARAM_USE_MI_BATCHBUFFER_START
+I915_SETPARAM_TEX_LRU_LOG_GRANULARITY
+I915_SETPARAM_ALLOW_BATCHBUFFER
+I915_SETPARAM_NUM_USED_FENCES
Signature · 9 lines
2.1.0


------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud 
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
#4 lore raw
In reply to Patrik Jakobsson
Quoted 15 lines · Patrik Jakobsson

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 14 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in

Yeah, this looks pretty cool to me. I'm not familiar with strace internals, but the split and extensible design looks reasonable; should make it easy to add other drivers and such in the future.

Who on the strace side can pick this up?

Thanks,
Jesse

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

#5 lore raw
Quoted 31 lines · Jesse Barnes
Quoted 15 lines · Patrik Jakobsson

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 14 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 5 lines

Yeah, this looks pretty cool to me. I'm not familiar with strace internals, but the split and extensible design looks reasonable; should make it easy to add other drivers and such in the future.

Who on the strace side can pick this up?

Hi Dmitry

Can I bother you with a review?

Thanks
Patrik

Quoted 4 lines

Thanks,
Jesse

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

#6 lore raw
In reply to Patrik Jakobsson
Quoted 15 lines · On Wed, 6 May 2015 16:48:01 +0200 Patrik Jakobsson…

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.
* This does not compile on my system (archlinux), because drm.h lives in libdrm/drm.h and not in drm/drm.h, I know it is an rfc, but this needs to use pkg-config in order to know where libdrm headers are.

#include <fcntl.h>
#include <sys/ioctl.h>
#include <unistd.h>

#include <drm.h>
#include <i915_drm.h>

int main()

{
	int fd = open("/dev/dri/card0", O_RDWR);
	ioctl(fd, DRM_IOW(DRM_COMMAND_BASE + DRM_I915_FLUSH, int), 0);
	return 0;
}
$ CFLAGS=-I/usr/include/libdrm make i915-ioctls
cc -I/usr/include/libdrm    i915-ioctls.c   -o i915-ioctls
$ sudo strace -e ioctl ./i915-ioctls 
ioctl(3, _IOC(_IOC_WRITE, 0x64, 0x41, 0x04), 0) = 0
+++ exited with 0 +++
$ sudo ~/source/strace/strace -e ioctl ./i915-ioctls 
ioctl(3, 0x40046441 /* I915_IOCTL_??? */, 0) = 0
+++ exited with 0 +++
Signature · 5 lines
Gabriel Laskar
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#7 lore raw
Quoted 34 lines · Gabriel Laskar
Quoted 15 lines · On Wed, 6 May 2015 16:48:01 +0200 Patrik Jakobsson…

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 2 lines

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Thanks for the review. Yes, my plan is to add generic drm ioctls as well.

Quoted 13 lines

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.

Yes, moving it to ioctl_decode_command_number() makes sense. I'll do that. And I'll make the output consistent and skip the I915_IOCTL_???. It comes with the drawback of possibly duplicated entries when doing the lookup even though we know we're talking to i915, but it is still nicer than _???.

Quoted 3 lines

* This does not compile on my system (archlinux), because drm.h lives in libdrm/drm.h and not in drm/drm.h, I know it is an rfc, but this needs to use pkg-config in order to know where libdrm headers are.

In theory libdrm is not needed since the headers are available in uapi, so I tried to avoid adding an additional dependency. But since distros do not guarantee the existance of <drm/drm.h> it might be better to use libdrm. I'll look into that.

Thanks
Patrik

Quoted 16 lines

#include <fcntl.h>
#include <sys/ioctl.h>
#include <unistd.h>

#include <drm.h>
#include <i915_drm.h>

int main()

{
	int fd = open("/dev/dri/card0", O_RDWR);
	ioctl(fd, DRM_IOW(DRM_COMMAND_BASE + DRM_I915_FLUSH, int), 0);
	return 0;
}
Quoted 12 lines
$ CFLAGS=-I/usr/include/libdrm make i915-ioctls
cc -I/usr/include/libdrm    i915-ioctls.c   -o i915-ioctls
$ sudo strace -e ioctl ./i915-ioctls 
ioctl(3, _IOC(_IOC_WRITE, 0x64, 0x41, 0x04), 0) = 0
+++ exited with 0 +++
$ sudo ~/source/strace/strace -e ioctl ./i915-ioctls 
ioctl(3, 0x40046441 /* I915_IOCTL_??? */, 0) = 0
+++ exited with 0 +++
Signature · 1 line
Gabriel Laskar

------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud Widest out-of-the-box monitoring support with 50+ applications Performance metrics, stats and reports that give you Actionable Insights Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
Strace-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/strace-devel

#8 lore raw
Quoted 54 lines · On Mon, 11 May 2015 15:54:24 +0200 Patrik Jakobsson…
Quoted 34 lines · Gabriel Laskar
Quoted 15 lines · On Wed, 6 May 2015 16:48:01 +0200 Patrik Jakobsson…

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 2 lines

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Thanks for the review. Yes, my plan is to add generic drm ioctls as well.

Quoted 13 lines

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.

Quoted 4 lines

Yes, moving it to ioctl_decode_command_number() makes sense. I'll do that. And I'll make the output consistent and skip the I915_IOCTL_???. It comes with the drawback of possibly duplicated entries when doing the lookup even though we know we're talking to i915, but it is still nicer than _???.

If you call all the request decoding code from ioctl_decode_command_number() you will still be able to determine if you are on i915, and write the correct request, but the fallback code will be no longer duplicated. You will have to call xlookup() instead of printxval(), in order to be able to know when the decoding fail though.

Quoted 8 lines
Quoted 3 lines

* This does not compile on my system (archlinux), because drm.h lives in libdrm/drm.h and not in drm/drm.h, I know it is an rfc, but this needs to use pkg-config in order to know where libdrm headers are.

In theory libdrm is not needed since the headers are available in uapi, so I tried to avoid adding an additional dependency. But since distros do not guarantee the existance of <drm/drm.h> it might be better to use libdrm. I'll look into that.

It seems that on archlinux, they remove intentionally[1] the headers for drm. For others, I can't say.

[1]: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/linux-api-headers#n40

Signature · 5 lines
Gabriel Laskar
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#9 lore raw
Quoted 62 lines · Gabriel Laskar
Quoted 54 lines · On Mon, 11 May 2015 15:54:24 +0200 Patrik Jakobsson…
Quoted 34 lines · Gabriel Laskar
Quoted 15 lines · On Wed, 6 May 2015 16:48:01 +0200 Patrik Jakobsson…

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 2 lines

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Thanks for the review. Yes, my plan is to add generic drm ioctls as well.

Quoted 13 lines

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.

Quoted 4 lines

Yes, moving it to ioctl_decode_command_number() makes sense. I'll do that. And I'll make the output consistent and skip the I915_IOCTL_???. It comes with the drawback of possibly duplicated entries when doing the lookup even though we know we're talking to i915, but it is still nicer than _???.

Quoted 5 lines

If you call all the request decoding code from ioctl_decode_command_number() you will still be able to determine if you are on i915, and write the correct request, but the fallback code will be no longer duplicated. You will have to call xlookup() instead of printxval(), in order to be able to know when the decoding fail though.

Unfortunately I cannot check for i915 in _decode_command_number since I don't have the full tcb context (cannot figure out the path, etc.). That's probably why I had to duplicate the decoding in the first place. The only other solution I see is to detect i915 early and store it globally somewhere but that is rather nasty. Not sure how to do this in a better way. What am I missing?

Quoted 10 lines
Quoted 9 lines
Quoted 4 lines

* This does not compile on my system (archlinux), because drm.h lives in libdrm/drm.h and not in drm/drm.h, I know it is an rfc, but this needs to use pkg-config in order to know where libdrm headers are.

In theory libdrm is not needed since the headers are available in uapi, so I tried to avoid adding an additional dependency. But since distros do not guarantee the existance of <drm/drm.h> it might be better to use libdrm. I'll look into that.

Quoted 4 lines

It seems that on archlinux, they remove intentionally[1] the headers for drm. For others, I can't say.

[1]: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/linux-api-headers#n40

Yes, it's prefered to use the headers from libdrm. It's actually not a bad thing that Arch Linux enforces this. I'll just have to add libdrm as a dependency for drm ioctl decoding.

Quoted 3 lines
Signature · 2 lines
-- 
Gabriel Laskar

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

#10 lore raw
Quoted 69 lines · On Tue, 12 May 2015 14:35:28 +0200 Patrik Jakobsson…
Quoted 62 lines · Gabriel Laskar
Quoted 54 lines · On Mon, 11 May 2015 15:54:24 +0200 Patrik Jakobsson…
Quoted 34 lines · Gabriel Laskar
Quoted 15 lines · On Wed, 6 May 2015 16:48:01 +0200 Patrik Jakobsson…

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 2 lines

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Thanks for the review. Yes, my plan is to add generic drm ioctls as well.

Quoted 13 lines

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.

Quoted 4 lines

Yes, moving it to ioctl_decode_command_number() makes sense. I'll do that. And I'll make the output consistent and skip the I915_IOCTL_???. It comes with the drawback of possibly duplicated entries when doing the lookup even though we know we're talking to i915, but it is still nicer than _???.

Quoted 5 lines

If you call all the request decoding code from ioctl_decode_command_number() you will still be able to determine if you are on i915, and write the correct request, but the fallback code will be no longer duplicated. You will have to call xlookup() instead of printxval(), in order to be able to know when the decoding fail though.

Quoted 5 lines

Unfortunately I cannot check for i915 in _decode_command_number since I don't have the full tcb context (cannot figure out the path, etc.). That's probably why I had to duplicate the decoding in the first place. The only other solution I see is to detect i915 early and store it globally somewhere but that is rather nasty. Not sure how to do this in a better way. What am I missing?

Oh yes, I see. We can add tcb context to ioctl_decode_command_number(), it should not be a problem. We didn't have the need for it for the moment, that's all.

Dmitry? What do you think of this?

Signature · 12 lines
Gabriel Laskar

------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud 
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
Strace-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/strace-devel

A message that is not in the archive (sent off-list or not yet received): 20150512193759.09bb505b-krIL5v34lyW+8jMViQwUxmazZaUMDOZU@public.gmane.org

#12 lore raw
Quoted 73 lines · Gabriel Laskar
Quoted 66 lines · Patrik Jakobsson
Quoted 59 lines · Gabriel Laskar
Quoted 53 lines · Patrik Jakobsson
Quoted 33 lines · Gabriel Laskar
Quoted 15 lines · Patrik Jakobsson

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 2 lines

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Thanks for the review. Yes, my plan is to add generic drm ioctls as well.

Quoted 13 lines

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.

Quoted 4 lines

Yes, moving it to ioctl_decode_command_number() makes sense. I'll do that. And I'll make the output consistent and skip the I915_IOCTL_???. It comes with the drawback of possibly duplicated entries when doing the lookup even though we know we're talking to i915, but it is still nicer than _???.

Quoted 5 lines

If you call all the request decoding code from ioctl_decode_command_number() you will still be able to determine if you are on i915, and write the correct request, but the fallback code will be no longer duplicated. You will have to call xlookup() instead of printxval(), in order to be able to know when the decoding fail though.

Quoted 5 lines

Unfortunately I cannot check for i915 in _decode_command_number since I don't have the full tcb context (cannot figure out the path, etc.). That's probably why I had to duplicate the decoding in the first place. The only other solution I see is to detect i915 early and store it globally somewhere but that is rather nasty. Not sure how to do this in a better way. What am I missing?

Quoted 5 lines

Oh yes, I see. We can add tcb context to ioctl_decode_command_number(), it should not be a problem. We didn't have the need for it for the moment, that's all.

Dmitry? What do you think of this?

I had no chance to have a look at the code, but anyway, passing tcb context down to ioctl_decode_command_number() shouldn't be a problem at all.

Signature · 12 lines
ldv

------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud 
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
Strace-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/strace-devel
#13 lore raw
Quoted 78 lines · Dmitry V. Levin
Quoted 77 lines · Gabriel Laskar
Quoted 66 lines · Patrik Jakobsson
Quoted 59 lines · Gabriel Laskar
Quoted 53 lines · Patrik Jakobsson
Quoted 33 lines · Gabriel Laskar
Quoted 15 lines · Patrik Jakobsson

This patch set aims to make strace more useful when tracing i915 ioctls. The ioctl type is first checked for being drm and then the driver backing the opened device is identified by looking at sysfs. Other drivers than i915 can easily be added.

Only a subset of the i915 ioctls are included. I will extend this patch set if the approach looks ok. The generic drm ioctls are also missing.

Give it a spin with:
strace -e trace=ioctl -p pidof X

Patrik Jakobsson (2):
strace/drm: Print extended info for drm and i915 ioctls strace/drm: Print args for most common i915 ioctls

Quoted 15 lines
 Makefile.am                |   2 +
 defs.h                     |   2 +
 drm.c                      | 104 +++++++++++++++++
 drm_i915.c                 | 278 +++++++++++++++++++++++++++++++++++++++++++++
 ioctl.c                    |   5 +
 xlat/drm_i915_getparams.in |  28 +++++
 xlat/drm_i915_ioctls.in    |  51 +++++++++
 xlat/drm_i915_setparams.in |   4 +
 8 files changed, 474 insertions(+)
 create mode 100644 drm.c
 create mode 100644 drm_i915.c
 create mode 100644 xlat/drm_i915_getparams.in
 create mode 100644 xlat/drm_i915_ioctls.in
 create mode 100644 xlat/drm_i915_setparams.in
Quoted 2 lines

This is a great start! We need this kind of decoding. Do you plan to add also the generic drm ioctl decoding?

Thanks for the review. Yes, my plan is to add generic drm ioctls as well.

Quoted 13 lines

Some issues though:

* The way you avoid the ioctl request decoding is quite hard to follow, but it seems that you don't have much of a choice, except that in drm_ioctl(), the code from SYS_FUNC(ioctl) is duplicated. It seems it needs some work here, to allow a simpler code path. Maybe this would be clearer if the decoding/drm_get_driver_name, etc… was in ioctl_decode_command_number(). Also, with the actual code, if you are on i915 with an invalid ioctl number, it will be printed as "I915_IOCTL_???" and not "_IOC(...)" (see below for an example.) This will also add an inconsistent result depending whether /sys is mounted or not.

Quoted 4 lines

Yes, moving it to ioctl_decode_command_number() makes sense. I'll do that. And I'll make the output consistent and skip the I915_IOCTL_???. It comes with the drawback of possibly duplicated entries when doing the lookup even though we know we're talking to i915, but it is still nicer than _???.

Quoted 5 lines

If you call all the request decoding code from ioctl_decode_command_number() you will still be able to determine if you are on i915, and write the correct request, but the fallback code will be no longer duplicated. You will have to call xlookup() instead of printxval(), in order to be able to know when the decoding fail though.

Quoted 5 lines

Unfortunately I cannot check for i915 in _decode_command_number since I don't have the full tcb context (cannot figure out the path, etc.). That's probably why I had to duplicate the decoding in the first place. The only other solution I see is to detect i915 early and store it globally somewhere but that is rather nasty. Not sure how to do this in a better way. What am I missing?

Oh yes, I see. We can add tcb context to ioctl_decode_command_number(), it should not be a problem. We didn't have the need for it for the moment, that's all.

Quoted 5 lines
Quoted 1 line

Dmitry? What do you think of this?

I had no chance to have a look at the code, but anyway, passing tcb context down to ioctl_decode_command_number() shouldn't be a problem at all.

Sounds good. Then I'll use ioctl_decode_command_number as the dispatcher.

Thanks
Patrik

Quoted 4 lines
Signature · 2 lines
-- 
ldv

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