Skip to contentSkip to conversation

igt_kms: Avoid NULL ptr deref when commiting disabled planes

PATCH In review Tvrtko Ursulin 2 participants 18 messages May 5, 2015 – May 8, 2015 Library code

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/4 igt_kms: Avoid NULL ptr deref when commiting disabled planes

From: Tvrtko Ursulin <[email protected]>

I think;

commit a26f9f9ad0e679c7ce413a25d34f6914e1174151
Author: chandra konduru <[email protected]>
Date: Mon Mar 30 13:52:04 2015 -0700

i-g-t: Adding plane scaling test case

introduced a condition where it attempts to update a disabled plane because of the newly introduced size_changed flag which is set for disabled frame buffers. Result is a NULL ptr deref in igt_drm_plane_commit (plane->fb->src_x).

Start recognising this case as disabled plane and act accordingly.

v2: Split out igt_plane_set_fb cleanup. (Thomas Wood)

Cc chandra konduru <[email protected]>
Cc Thomas Wood <[email protected]>

---
There might be a better fix, but this works for me.

---
 lib/igt_kms.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/igt_kms.c +1 -1
index b7d1e90..33d437d 100644
@@ -1331,7 +1331,7 @@ static int igt_drm_plane_commit(igt_plane_t *plane,
 	fb_id = igt_plane_get_fb_id(plane);
 	crtc_id = output->config.crtc->crtc_id;
 
-	if (plane->fb_changed && fb_id == 0) {
+	if ((plane->fb_changed || plane->size_changed) && fb_id == 0) {
 		LOG(display,
 		    "%s: SetPlane pipe %s, plane %d, disabling\n",
 		    igt_output_name(output),
Signature · 6 lines
2.3.5

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

PATCH 2/4 igt_kms: Merge condition in igt_plane_set_fb

From: Tvrtko Ursulin <[email protected]>

There were two paths for fb and !fb.

Cc chandra konduru <[email protected]>
Cc Thomas Wood <[email protected]>
---
 lib/igt_kms.c | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)
lib/igt_kms.c +3 -8
index 33d437d..b5ba273 100644
@@ -1765,14 +1765,6 @@ void igt_plane_set_fb(igt_plane_t *plane, struct igt_fb *fb)
 	plane->fb = fb;
 	/* hack to keep tests working that don't call igt_plane_set_size() */
 	if (fb) {
-		plane->crtc_w = fb->width;
-		plane->crtc_h = fb->height;
-	} else {
-		plane->crtc_w = 0;
-		plane->crtc_h = 0;
-	}
-
-	if (fb) {
 		/* set default plane pos/size as fb size */
 		plane->crtc_x = 0;
 		plane->crtc_y = 0;
@@ -1784,6 +1776,9 @@ void igt_plane_set_fb(igt_plane_t *plane, struct igt_fb *fb)
 		fb->src_y = 0;
 		fb->src_w = fb->width;
 		fb->src_h = fb->height;
+	} else {
+		plane->crtc_w = 0;
+		plane->crtc_h = 0;
 	}
 
 	plane->fb_changed = true;
Signature · 6 lines
2.3.5

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

Re: [PATCH i-g-t 2/4] igt_kms: Merge condition in igt_plane_set_fb

Quoted 19 lines
lib/igt_kms.c +0 -8
@@ -1765,14 +1765,6 @@ void igt_plane_set_fb(igt_plane_t *plane, struct
 igt_fb *fb)
 	plane->fb = fb;
 	/* hack to keep tests working that don't call igt_plane_set_size() */
 	if (fb) {
-		plane->crtc_w = fb->width;
-		plane->crtc_h = fb->height;
-	} else {
-		plane->crtc_w = 0;
-		plane->crtc_h = 0;
-	}
-
-	if (fb) {
 		/* set default plane pos/size as fb size */
 		plane->crtc_x = 0;
lib/igt_kms.c +0 -0
 		plane->crtc_y = 0;

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, struct · line 1783

Quoted 8 lines
lib/igt_kms.c +3 -0
@@ -1784,6 +1776,9 @@ void igt_plane_set_fb(igt_plane_t *plane, struct
 igt_fb *fb)
 		fb->src_y = 0;
 		fb->src_w = fb->width;
 		fb->src_h = fb->height;
+	} else {
+		plane->crtc_w = 0;
+		plane->crtc_h = 0;
 	}

Existing code is simply setting fb src position and plane crtc position to 0s (top left) and src size as fb size and crtc size as plane size to start a fb with a plane. Then individual test can change them to whatever fb position/size and plane position/size as it wants. As I commented to 3/4 patch, if these initializations are removed, then all tests to be updated to explicitly set them.

As a side note, is there any reason for having two patches 2/4 and 3/4 modifying same lines of code instead of a single patch?

Quoted 4 lines
(quoted diff) +0 -1
 	plane->fb_changed = true;
--

2.3.5

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

#3 lore raw

Re: [PATCH i-g-t 2/4] igt_kms: Merge condition in igt_plane_set_fb

Quoted 22 lines · Chandra
Quoted 22 lines
lib/igt_kms.c +0 -8
@@ -1765,14 +1765,6 @@ void igt_plane_set_fb(igt_plane_t *plane, struct
 igt_fb *fb)
  	plane->fb = fb;
  	/* hack to keep tests working that don't call igt_plane_set_size() */
  	if (fb) {
-		plane->crtc_w = fb->width;
-		plane->crtc_h = fb->height;
-	} else {
-		plane->crtc_w = 0;
-		plane->crtc_h = 0;
-	}
-
-	if (fb) {
  		/* set default plane pos/size as fb size */
  		plane->crtc_x = 0;
  		plane->crtc_y = 0;
lib/igt_kms.c +1 -0
@@ -1784,6 +1776,9 @@ void igt_plane_set_fb(igt_plane_t *plane, struct
 igt_fb *fb)
  		fb->src_y = 0;
  		fb->src_w = fb->width;
  		fb->src_h = fb->height;
+	} else {
Quoted 8 lines

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, struct igt_fb *fb) · line 1782

Quoted 3 lines
lib/igt_kms.c +2 -0
+		plane->crtc_w = 0;
+		plane->crtc_h = 0;
  	}

Existing code is simply setting fb src position and plane crtc position to 0s (top left) and src size as fb size and crtc size as plane size to start a fb with a plane. Then individual test can change them to whatever fb position/size and plane position/size as it wants. As I commented to 3/4 patch, if these initializations are removed, then all tests to be updated to explicitly set them.

Not sure what you mean. I simply cleaned two "if (fb)" conditions one after another, into one. No functional changes.

Quoted 2 lines

As a side note, is there any reason for having two patches 2/4 and 3/4 modifying same lines of code instead of a single patch?

Because this is just a code cleanup and the other was a functional change. And because it doesn't matter - lets not spend hours going back and forth on trivial IGT fixes.

Regards,

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

#4 lore raw

PATCH 3/4 igt_kms: Do not reset plane position on assigning a fb

In reply to Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

commit a26f9f9ad0e679c7ce413a25d34f6914e1174151
Author: chandra konduru <[email protected]>
Date: Mon Mar 30 13:52:04 2015 -0700

i-g-t: Adding plane scaling test case

Started doing this and broke kms_rotation_crc.

---
 lib/igt_kms.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
lib/igt_kms.c +1 -3
index b5ba273..0665d70 100644
@@ -1765,9 +1765,7 @@ void igt_plane_set_fb(igt_plane_t *plane, struct igt_fb *fb)
 	plane->fb = fb;
 	/* hack to keep tests working that don't call igt_plane_set_size() */
 	if (fb) {
-		/* set default plane pos/size as fb size */
-		plane->crtc_x = 0;
-		plane->crtc_y = 0;
+		/* set default plane size as fb size */
 		plane->crtc_w = fb->width;
 		plane->crtc_h = fb->height;
 
Signature · 6 lines
2.3.5

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

Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, struct · line 1771

Quoted 12 lines
lib/igt_kms.c +0 -3
@@ -1765,9 +1765,7 @@ void igt_plane_set_fb(igt_plane_t *plane, struct
 igt_fb *fb)
 	plane->fb = fb;
 	/* hack to keep tests working that don't call igt_plane_set_size() */
 	if (fb) {
-		/* set default plane pos/size as fb size */
-		plane->crtc_x = 0;
-		plane->crtc_y = 0;

Reason for doing this is to make existing kms tests to run without modification. Otherwise every existing test has to explicitly call igt_plane_set_position to make sure it works. Can you make sure removing above 2 lines doesn't break existing tests?

By the way, instead of this, did you look at the option of changing kms_rotation_crc to set plane position to whatever required when fb is changed.

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

#6 lore raw

Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 15 lines · Chandra

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, struct · line 1771

Quoted 12 lines
lib/igt_kms.c +0 -3
@@ -1765,9 +1765,7 @@ void igt_plane_set_fb(igt_plane_t *plane, struct
 igt_fb *fb)
  	plane->fb = fb;
  	/* hack to keep tests working that don't call igt_plane_set_size() */
  	if (fb) {
-		/* set default plane pos/size as fb size */
-		plane->crtc_x = 0;
-		plane->crtc_y = 0;

Reason for doing this is to make existing kms tests to run without modification. Otherwise every existing test has to explicitly call igt_plane_set_position to make sure it works. Can you make sure removing above 2 lines doesn't break existing tests?

Hm, before your patch tests could set a plane position, change the fb, and plane position would remain the same. After your patch changing the fb resets the plane position. My patch reverts this behaviour to old one.

So I am not sure in what cases resetting plane position on fb changes helps and which tests?

Which tests do you think should be run?

Regards,

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

#7 lore raw

Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 31 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Thursday, May 07, 2015 2:15 AM
To: Konduru, Chandra; [email protected]
Cc: Ursulin, Tvrtko
Subject: Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 14 lines · Chandra

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, · line 1771

Quoted 13 lines
lib/igt_kms.c index b5ba273..0665d70 renamed from lib/igt_kms.c +0 -0

100644

lib/igt_kms.c +0 -3
@@ -1765,9 +1765,7 @@ void igt_plane_set_fb(igt_plane_t *plane,
 struct igt_fb *fb)
  	plane->fb = fb;
  	/* hack to keep tests working that don't call igt_plane_set_size() */
  	if (fb) {
-		/* set default plane pos/size as fb size */
-		plane->crtc_x = 0;
-		plane->crtc_y = 0;

Reason for doing this is to make existing kms tests to run without

modification.

Quoted 2 lines

Otherwise every existing test has to explicitly call igt_plane_set_position to make sure it works. Can you make sure removing

above 2 lines doesn't break existing tests?

Quoted 9 lines

Hm, before your patch tests could set a plane position, change the fb, and plane position would remain the same. After your patch changing the fb resets the plane position. My patch reverts this behaviour to old one.

So I am not sure in what cases resetting plane position on fb changes helps and which tests?

Which tests do you think should be run?

Ignore my previous comment, I was thinking that this change will make plane->crtc_x/y uninitialized before calling igt_drm_plane_commit. But igt_display_init() does memsets display which will initialize these to 0s. So no initialization required as part here. Your changes are fine.

Quoted 3 lines

Regards,

Tvrtko

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

#8 lore raw

Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 36 lines · Chandra
Quoted 36 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Thursday, May 07, 2015 2:15 AM
To: Konduru, Chandra; [email protected]
Cc: Ursulin, Tvrtko
Subject: Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 14 lines · Chandra

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, · line 1771

Quoted 13 lines
lib/igt_kms.c index b5ba273..0665d70 renamed from lib/igt_kms.c +0 -0

100644

lib/igt_kms.c +0 -3
@@ -1765,9 +1765,7 @@ void igt_plane_set_fb(igt_plane_t *plane,
 struct igt_fb *fb)
   	plane->fb = fb;
   	/* hack to keep tests working that don't call igt_plane_set_size() */
   	if (fb) {
-		/* set default plane pos/size as fb size */
-		plane->crtc_x = 0;
-		plane->crtc_y = 0;

Reason for doing this is to make existing kms tests to run without

modification.

Quoted 2 lines

Otherwise every existing test has to explicitly call igt_plane_set_position to make sure it works. Can you make sure removing

above 2 lines doesn't break existing tests?

Hm, before your patch tests could set a plane position, change the fb, and plane position would remain the same. After your patch changing the fb resets the plane position. My patch reverts this behaviour to old one.

So I am not sure in what cases resetting plane position on fb changes helps and which tests?

Quoted 6 lines
Quoted 2 lines

Which tests do you think should be run?

Ignore my previous comment, I was thinking that this change will make plane->crtc_x/y uninitialized before calling igt_drm_plane_commit. But igt_display_init() does memsets display which will initialize these to 0s. So no initialization required as part here. Your changes are fine.

So r-b on patches 1-3 ?

Regards,

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

Reviewed-by
#9 lore raw

Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 51 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Friday, May 08, 2015 1:36 AM
To: Konduru, Chandra; [email protected]
Cc: Ursulin, Tvrtko
Subject: Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 25 lines · Chandra
Quoted 25 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Thursday, May 07, 2015 2:15 AM
To: Konduru, Chandra; [email protected]
Cc: Ursulin, Tvrtko
Subject: Re: [PATCH i-g-t 3/4] igt_kms: Do not reset plane position on assigning a fb

Quoted 14 lines · Chandra

lib/igt_kms.c void igt_plane_set_fb(igt_plane_t *plane, · line 1771

Quoted 13 lines
lib/igt_kms.c index b5ba273..0665d70 renamed from lib/igt_kms.c +0 -0

100644

lib/igt_kms.c +0 -3
@@ -1765,9 +1765,7 @@ void igt_plane_set_fb(igt_plane_t *plane,
 struct igt_fb *fb)
   	plane->fb = fb;
   	/* hack to keep tests working that don't call igt_plane_set_size() */
   	if (fb) {
-		/* set default plane pos/size as fb size */
-		plane->crtc_x = 0;
-		plane->crtc_y = 0;

Reason for doing this is to make existing kms tests to run without

modification.

Quoted 8 lines
Quoted 8 lines
Quoted 3 lines

Otherwise every existing test has to explicitly call igt_plane_set_position to make sure it works. Can you make sure removing

above 2 lines doesn't break existing tests?

Hm, before your patch tests could set a plane position, change the fb, and plane position would remain the same. After your patch changing the fb resets the plane position. My patch reverts this behaviour to

old one.

Quoted 4 lines
Quoted 4 lines

So I am not sure in what cases resetting plane position on fb changes helps and which tests?

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

Which tests do you think should be run?

Ignore my previous comment, I was thinking that this change will make plane->crtc_x/y uninitialized before calling igt_drm_plane_commit. But igt_display_init() does memsets display which will initialize these to 0s. So no initialization required as part here. Your changes are fine.

So r-b on patches 1-3 ?

For patches 1-3:

Quoted 4 lines

Regards,

Tvrtko

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

#10 lore raw

PATCH 4/4 kms_plane_scaling: Find the image regardless how the test is run

In reply to Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

As it stands running the test like "sudo tests/kms_plane_scaling" does not work.

Fix it by using the same method igt_paint_image uses.

v2: Export Cairo read callback instead of duplicating it. (Thomas Wood)

Cc chandra konduru <[email protected]>
Cc Thomas Wood <[email protected]>
---
 lib/igt_fb.c              | 17 ++++++++++++++---
 lib/igt_fb.h              |  2 ++
 tests/kms_plane_scaling.c |  7 ++++++-
 3 files changed, 22 insertions(+), 4 deletions(-)
lib/igt_fb.c +14 -3
index cc4b8ee..2416e76 100644
@@ -338,8 +338,18 @@ void igt_paint_test_pattern(cairo_t *cr, int width, int height)
 	igt_assert(!cairo_status(cr));
 }
 
-static cairo_status_t
-stdio_read_func(void *closure, unsigned char* data, unsigned int size)
+/**
+ * igt_cairo_read_func:
+ * @closure: callback closure parameter
+ * @data: callback data parameter
+ * @size: callback size parameter
+ *
+ * Read callback for cairo_image_surface_create_from_png_stream to be used
+ * in conjuction with igt_fopen_data in order to open data files from IGT
+ * standard locations.
+ */
+cairo_status_t
+igt_cairo_read_func(void *closure, unsigned char* data, unsigned int size)
 {
 	if (fread(data, 1, size, (FILE*)closure) != size)
 		return CAIRO_STATUS_READ_ERROR;
@@ -369,7 +379,8 @@ void igt_paint_image(cairo_t *cr, const char *filename,
 
 	f = igt_fopen_data(filename);
 
-	image = cairo_image_surface_create_from_png_stream(&stdio_read_func, f);
+	image = cairo_image_surface_create_from_png_stream(&igt_cairo_read_func,
+							   f);
 	igt_assert(cairo_surface_status(image) == CAIRO_STATUS_SUCCESS);
 
 	img_width = cairo_image_surface_get_width(image);
lib/igt_fb.h +2 -0
index a07acd2..2da5f0c 100644
@@ -98,6 +98,8 @@ void igt_write_fb_to_png(int fd, struct igt_fb *fb, const char *filename);
 int igt_cairo_printf_line(cairo_t *cr, enum igt_text_align align,
 			       double yspacing, const char *fmt, ...)
 			       __attribute__((format (printf, 4, 5)));
+cairo_status_t
+igt_cairo_read_func(void *closure, unsigned char* data, unsigned int size);
 
 /* helpers to handle drm fourcc codes */
 uint32_t igt_bpp_depth_to_drm_format(int bpp, int depth);
tests/kms_plane_scaling.c +6 -1
index 00db5cb..272f759 100644
@@ -210,6 +210,7 @@ static void test_plane_scaling(data_t *d)
 	enum pipe pipe;
 	int valid_tests = 0;
 	int primary_plane_scaling = 0; /* For now */
+	FILE* f;
 
 	igt_require(d->display.has_universal_planes);
 	igt_require(d->num_scalers);
@@ -223,11 +224,15 @@ static void test_plane_scaling(data_t *d)
 		mode = igt_output_get_mode(output);
 
 		/* allocate fb2 with image size */
-		image = cairo_image_surface_create_from_png(FILE_NAME);
+
+		f = igt_fopen_data(FILE_NAME);
+		image = cairo_image_surface_create_from_png_stream(
+						&igt_cairo_read_func, f);
 		igt_assert(cairo_surface_status(image) == CAIRO_STATUS_SUCCESS);
 		d->image_w = cairo_image_surface_get_width(image);
 		d->image_h = cairo_image_surface_get_height(image);
 		cairo_surface_destroy(image);
+		fclose(f);
 
 		d->fb_id2 = igt_create_fb(d->drm_fd,
 				d->image_w, d->image_h,
Signature · 6 lines
2.3.5

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

Re: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 10 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Tuesday, May 05, 2015 2:53 AM
To: [email protected]
Cc: Ursulin, Tvrtko; Konduru, Chandra; Wood, Thomas
Subject: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 6 lines

From: Tvrtko Ursulin <[email protected]>

As it stands running the test like "sudo tests/kms_plane_scaling" does not work.

Fix it by using the same method igt_paint_image uses.

As image size is required in other tests too, in recent patch I added igt_get_image_size() which can be used here instead duplication.

Quoted 82 lines

v2: Export Cairo read callback instead of duplicating it. (Thomas Wood)

Cc chandra konduru <[email protected]>
Cc Thomas Wood <[email protected]>
---
 lib/igt_fb.c              | 17 ++++++++++++++---
 lib/igt_fb.h              |  2 ++
 tests/kms_plane_scaling.c |  7 ++++++-
 3 files changed, 22 insertions(+), 4 deletions(-)
lib/igt_fb.c +17 -3
@@ -338,8 +338,18 @@ void igt_paint_test_pattern(cairo_t *cr, int width, int
 height)
 	igt_assert(!cairo_status(cr));
 }
 
-static cairo_status_t
-stdio_read_func(void *closure, unsigned char* data, unsigned int size)
+/**
+ * igt_cairo_read_func:
+ * @closure: callback closure parameter
+ * @data: callback data parameter
+ * @size: callback size parameter
+ *
+ * Read callback for cairo_image_surface_create_from_png_stream to be
+used
+ * in conjuction with igt_fopen_data in order to open data files from
+IGT
+ * standard locations.
+ */
+cairo_status_t
+igt_cairo_read_func(void *closure, unsigned char* data, unsigned int
+size)
 {
 	if (fread(data, 1, size, (FILE*)closure) != size)
 		return CAIRO_STATUS_READ_ERROR;
@@ -369,7 +379,8 @@ void igt_paint_image(cairo_t *cr, const char *filename,
 
 	f = igt_fopen_data(filename);
 
-	image =
 cairo_image_surface_create_from_png_stream(&stdio_read_func, f);
+	image =
 cairo_image_surface_create_from_png_stream(&igt_cairo_read_func,
+							   f);
 	igt_assert(cairo_surface_status(image) == CAIRO_STATUS_SUCCESS);
lib/igt_fb.c +0 -0
 	img_width = cairo_image_surface_get_width(image);
lib/igt_fb.h +3 -0
@@ -98,6 +98,8 @@ void igt_write_fb_to_png(int fd, struct igt_fb *fb, const
 char *filename);  int igt_cairo_printf_line(cairo_t *cr, enum igt_text_align align,
 			       double yspacing, const char *fmt, ...)
 			       __attribute__((format (printf, 4, 5)));
+cairo_status_t
+igt_cairo_read_func(void *closure, unsigned char* data, unsigned int
+size);
 
 /* helpers to handle drm fourcc codes */  uint32_t

igt_bpp_depth_to_drm_format(int bpp, int depth); diff --git a/tests/kms_plane_scaling.c b/tests/kms_plane_scaling.c index 00db5cb..272f759 100644

tests/kms_plane_scaling.c +1 -1
@@ -210,6 +210,7 @@ static void test_plane_scaling(data_t *d)
 	enum pipe pipe;
 	int valid_tests = 0;
 	int primary_plane_scaling = 0; /* For now */
+	FILE* f;
 
 	igt_require(d->display.has_universal_planes);
 	igt_require(d->num_scalers);
@@ -223,11 +224,15 @@ static void test_plane_scaling(data_t *d)
 		mode = igt_output_get_mode(output);
 
 		/* allocate fb2 with image size */
-		image = cairo_image_surface_create_from_png(FILE_NAME);

tests/kms_plane_scaling.c static void test_plane_scaling(data_t *d) · line 236

Quoted 10 lines
tests/kms_plane_scaling.c +5 -0
+
+		f = igt_fopen_data(FILE_NAME);
+		image = cairo_image_surface_create_from_png_stream(
+						&igt_cairo_read_func, f);
 		igt_assert(cairo_surface_status(image) ==
 CAIRO_STATUS_SUCCESS);
 		d->image_w = cairo_image_surface_get_width(image);
 		d->image_h = cairo_image_surface_get_height(image);
 		cairo_surface_destroy(image);
+		fclose(f);

Above can be replaced with

	igt_get_image_size(FILE_NAME, &d->image_w, &d->image_h);
Quoted 5 lines
tests/kms_plane_scaling.c +0 -1
 		d->fb_id2 = igt_create_fb(d->drm_fd,
 				d->image_w, d->image_h,
--

2.3.5

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

#12 lore raw

Re: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 10 lines · Chandra
Quoted 10 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Tuesday, May 05, 2015 2:53 AM
To: [email protected]
Cc: Ursulin, Tvrtko; Konduru, Chandra; Wood, Thomas
Subject: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

From: Tvrtko Ursulin <[email protected]>

Quoted 7 lines
Quoted 4 lines

As it stands running the test like "sudo tests/kms_plane_scaling" does not work.

Fix it by using the same method igt_paint_image uses.

As image size is required in other tests too, in recent patch I added igt_get_image_size() which can be used here instead duplication.

Ok, please do, I only attempted to fix it along the way since it did not run for me. And at the time there was no igt_get_image_size.

Regards,

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

#13 lore raw

Re: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 25 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Wednesday, May 06, 2015 2:29 AM
To: Konduru, Chandra; [email protected]
Cc: Ursulin, Tvrtko; Wood, Thomas
Subject: Re: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 13 lines · Chandra
Quoted 13 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Tuesday, May 05, 2015 2:53 AM
To: [email protected]
Cc: Ursulin, Tvrtko; Konduru, Chandra; Wood, Thomas
Subject: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

From: Tvrtko Ursulin <[email protected]>

As it stands running the test like "sudo tests/kms_plane_scaling" does not work.

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

Fix it by using the same method igt_paint_image uses.

As image size is required in other tests too, in recent patch I added igt_get_image_size() which can be used here instead duplication.

Ok, please do, I only attempted to fix it along the way since it did not run for me. And at the time there was no igt_get_image_size.

Yeah, you are right. Sure I have few things to changes to do anyway for adding rotation to kms_plane_scaling. Along with that, I will make above change. In that case, I presume you drop this patch.

Quoted 4 lines

Regards,

Tvrtko

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

#14 lore raw

Re: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 23 lines · Chandra
Quoted 23 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Wednesday, May 06, 2015 2:29 AM
To: Konduru, Chandra; [email protected]
Cc: Ursulin, Tvrtko; Wood, Thomas
Subject: Re: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

Quoted 13 lines · Chandra
Quoted 13 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Tuesday, May 05, 2015 2:53 AM
To: [email protected]
Cc: Ursulin, Tvrtko; Konduru, Chandra; Wood, Thomas
Subject: [PATCH i-g-t 4/4] kms_plane_scaling: Find the image regardless how the test is run

From: Tvrtko Ursulin <[email protected]>

As it stands running the test like "sudo tests/kms_plane_scaling" does not work.

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
Quoted 1 line

Fix it by using the same method igt_paint_image uses.

As image size is required in other tests too, in recent patch I added igt_get_image_size() which can be used here instead duplication.

Ok, please do, I only attempted to fix it along the way since it did not run for me. And at the time there was no igt_get_image_size.

Yeah, you are right. Sure I have few things to changes to do anyway for adding rotation to kms_plane_scaling. Along with that, I will make above change. In that case, I presume you drop this patch.

Dropping happily.

Regards,

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

#15 lore raw
In reply to Tvrtko Ursulin
Quoted 41 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Tuesday, May 05, 2015 2:53 AM
To: [email protected]
Cc: Ursulin, Tvrtko; Konduru, Chandra; Wood, Thomas
Subject: [PATCH v2 i-g-t 1/4] igt_kms: Avoid NULL ptr deref when commiting disabled planes

From: Tvrtko Ursulin <[email protected]>

I think;

commit a26f9f9ad0e679c7ce413a25d34f6914e1174151
Author: chandra konduru <[email protected]>
Date: Mon Mar 30 13:52:04 2015 -0700

i-g-t: Adding plane scaling test case

introduced a condition where it attempts to update a disabled plane because of the newly introduced size_changed flag which is set for disabled frame buffers. Result is a NULL ptr deref in igt_drm_plane_commit (plane->fb->src_x).

Start recognising this case as disabled plane and act accordingly.

v2: Split out igt_plane_set_fb cleanup. (Thomas Wood)

Cc chandra konduru <[email protected]>
Cc Thomas Wood <[email protected]>

---
There might be a better fix, but this works for me.

---
 lib/igt_kms.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

lib/igt_kms.c static int igt_drm_plane_commit(igt_plane_t *plane, · line 1334

Quoted 5 lines
lib/igt_kms.c +1 -1
@@ -1331,7 +1331,7 @@ static int igt_drm_plane_commit(igt_plane_t *plane,
 	fb_id = igt_plane_get_fb_id(plane);
 	crtc_id = output->config.crtc->crtc_id;
 
-	if (plane->fb_changed && fb_id == 0) {
+	if ((plane->fb_changed || plane->size_changed) && fb_id == 0) {

Shouldn't this include plane->position_changed too? Like:

	if ((plane->fb_changed || plane->size_changed || plane->position_changed) && fb_id == 0) {
Quoted 5 lines
lib/igt_kms.c +0 -1
 		LOG(display,
 		    "%s: SetPlane pipe %s, plane %d, disabling\n",
 		    igt_output_name(output),
--

2.3.5

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

#16 lore raw
Quoted 42 lines · Chandra
Quoted 42 lines

-----Original Message-----
From: Tvrtko Ursulin [mailto:[email protected]]
Sent: Tuesday, May 05, 2015 2:53 AM
To: [email protected]
Cc: Ursulin, Tvrtko; Konduru, Chandra; Wood, Thomas
Subject: [PATCH v2 i-g-t 1/4] igt_kms: Avoid NULL ptr deref when commiting disabled planes

From: Tvrtko Ursulin <[email protected]>

I think;

commit a26f9f9ad0e679c7ce413a25d34f6914e1174151
Author: chandra konduru <[email protected]>
Date: Mon Mar 30 13:52:04 2015 -0700

i-g-t: Adding plane scaling test case

introduced a condition where it attempts to update a disabled plane because of the newly introduced size_changed flag which is set for disabled frame buffers. Result is a NULL ptr deref in igt_drm_plane_commit (plane->fb->src_x).

Start recognising this case as disabled plane and act accordingly.

v2: Split out igt_plane_set_fb cleanup. (Thomas Wood)

Cc chandra konduru <[email protected]>
Cc Thomas Wood <[email protected]>

---
There might be a better fix, but this works for me.

---
  lib/igt_kms.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
lib/igt_kms.c +0 -0
@@ -1331,7 +1331,7 @@ static int igt_drm_plane_commit(igt_plane_t *plane,
  	fb_id = igt_plane_get_fb_id(plane);
  	crtc_id = output->config.crtc->crtc_id;
 
Quoted 5 lines

lib/igt_kms.c static int igt_drm_plane_commit(igt_plane_t *plane, · line 1334

Quoted 2 lines
lib/igt_kms.c +1 -1
-	if (plane->fb_changed && fb_id == 0) {
+	if ((plane->fb_changed || plane->size_changed) && fb_id == 0) {

Shouldn't this include plane->position_changed too? Like:

	if ((plane->fb_changed || plane->size_changed || plane->position_changed) && fb_id == 0) {

When you added size_changed, state for position_changed and fb == NULL remained the same, while size_changed added new state for size_changed == true and fb == NULL. So I added handling for that and did not think much beyond it. It fixes a segfault so I moved on. Or in other words, I don't see how it would harm to merge this, it doesn't make anything worse.

Regards,

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

#17 lore raw
Quoted 16 lines
Quoted 5 lines

lib/igt_kms.c static int igt_drm_plane_commit(igt_plane_t

Quoted 5 lines
lib/igt_kms.c index b7d1e90..33d437d renamed from lib/igt_kms.c +0 -0

100644

lib/igt_kms.c +0 -0
@@ -1331,7 +1331,7 @@ static int igt_drm_plane_commit(igt_plane_t

*plane,

Quoted 9 lines

lib/igt_kms.c static int igt_drm_plane_commit(igt_plane_t *plane, · line 1334

Quoted 5 lines
lib/igt_kms.c +1 -1
  	fb_id = igt_plane_get_fb_id(plane);
  	crtc_id = output->config.crtc->crtc_id;
 
-	if (plane->fb_changed && fb_id == 0) {
+	if ((plane->fb_changed || plane->size_changed) && fb_id == 0) {

Shouldn't this include plane->position_changed too? Like:

	if ((plane->fb_changed || plane->size_changed ||

plane->position_changed) && fb_id == 0) {

Quoted 5 lines

When you added size_changed, state for position_changed and fb == NULL remained the same, while size_changed added new state for size_changed == true and fb == NULL. So I added handling for that and did not think much beyond it. It fixes a segfault so I moved on. Or in other words, I don't see how it would harm to merge this, it doesn't make anything worse.

Agree it doesn't harm to merge.

Quoted 4 lines

Regards,

Tvrtko

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