Skip to contentSkip to conversation

drm/i915: Fix copy_to_user usage for pipe_crc

PATCH In review Rodrigo Vivi 5 participants 9 messages Aug 3, 2016 – Aug 5, 2016

Subsystems (2)
lore
Share & tools
lore

Rodrigo Vivi

author maintainer
#0 lore raw

PATCH 1/2 drm/i915: Fix copy_to_user usage for pipe_crc

Copy to user return the number of bytes it couldn't write and zero on success. So any number different than 0 should be considered a fault, not only when it doesn't write the full size.

---
 drivers/gpu/drm/i915/i915_debugfs.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +1 -3
index 7c42ec4..7052c47 100644
@@ -3632,7 +3632,6 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 	while (n_entries > 0) {
 		struct intel_pipe_crc_entry *entry =
 			&pipe_crc->entries[pipe_crc->tail];
-		int ret;
 
 		if (CIRC_CNT(pipe_crc->head, pipe_crc->tail,
 			     INTEL_PIPE_CRC_ENTRIES_NR) < 1)
@@ -3649,8 +3648,7 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 
 		spin_unlock_irq(&pipe_crc->lock);
 
-		ret = copy_to_user(user_buf, buf, PIPE_CRC_LINE_LEN);
-		if (ret == PIPE_CRC_LINE_LEN)
+		if (!copy_to_user(user_buf, buf, PIPE_CRC_LINE_LEN))
 			return -EFAULT;
 
 		user_buf += PIPE_CRC_LINE_LEN;
Signature · 6 lines
2.5.5

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

PATCH 2/2 drm/i915: Fix the return value of pipe crc read function.

A read(fd, buf, len) function should return the number of bytes read. In our case we need to return the number of bytes we copy to user, instead of returning the number of bytes we read internally.

It was really strange when I saw i-g-t test case using len '54' but getting '56' as return. First thought was how do we read more than we asked? But also I checked and there was really only 54. Until I realized it was all our fault. EFAULT!

---
 drivers/gpu/drm/i915/i915_debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_debugfs.c +1 -1
index 7052c47..b7b8d79 100644
@@ -3659,7 +3659,7 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 
 	spin_unlock_irq(&pipe_crc->lock);
 
-	return bytes_read;
+	return PIPE_CRC_LINE_LEN;
 }
 
 static const struct file_operations i915_pipe_crc_fops = {
Signature · 6 lines
2.5.5

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

Re: [PATCH 2/2] drm/i915: Fix the return value of pipe crc read function.

Quoted 21 lines · Rodrigo Vivi

A read(fd, buf, len) function should return the number of bytes read. In our case we need to return the number of bytes we copy to user, instead of returning the number of bytes we read internally.

It was really strange when I saw i-g-t test case using len '54' but getting '56' as return. First thought was how do we read more than we asked? But also I checked and there was really only 54. Until I realized it was all our fault. EFAULT!

---
 drivers/gpu/drm/i915/i915_debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
index 7052c47..b7b8d79 100644

drivers/gpu/drm/i915/i915_debugfs.c i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count, · line 3662

Quoted 5 lines
drivers/gpu/drm/i915/i915_debugfs.c +1 -1
@@ -3659,7 +3659,7 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 
 	spin_unlock_irq(&pipe_crc->lock);
 
-	return bytes_read;
+	return PIPE_CRC_LINE_LEN;

Nope. We can read multiple entries in one go.

bytes_read += snprintf()

so it's mostly good. The only case where it fails if we don't have enough space for the snprintf(), as snprintf() will then return the number of bytes it would have written. That can actually happen on account of the hex vs. decimal mess with the frame counter.

Quoted 10 lines
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
 }
 
 static const struct file_operations i915_pipe_crc_fops = {
Signature · 6 lines
2.5.5

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author maintainer
#3 lore raw

Re: [PATCH 2/2] drm/i915: Fix the return value of pipe crc read function.

Quoted 32 lines · Ville Syrjälä
Quoted 27 lines · Rodrigo Vivi

A read(fd, buf, len) function should return the number of bytes read. In our case we need to return the number of bytes we copy to user, instead of returning the number of bytes we read internally.

It was really strange when I saw i-g-t test case using len '54' but getting '56' as return. First thought was how do we read more than we asked? But also I checked and there was really only 54. Until I realized it was all our fault. EFAULT!

---
 drivers/gpu/drm/i915/i915_debugfs.c | 2 +-

 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
index 7052c47..b7b8d79 100644
@@ -3659,7 +3659,7 @@ i915_pipe_crc_read(struct file *filep, char
 __user *user_buf, size_t count,


  spin_unlock_irq(&pipe_crc->lock);

drivers/gpu/drm/i915/i915_debugfs.c i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count, · line 3662

Quoted 2 lines
drivers/gpu/drm/i915/i915_debugfs.c +1 -1
-	return bytes_read;
+	return PIPE_CRC_LINE_LEN;

Nope. We can read multiple entries in one go.

Quoted 6 lines

bytes_read += snprintf()

so it's mostly good. The only case where it fails if we don't have enough space for the snprintf(), as snprintf() will then return the number of bytes it would have written. That can actually happen on account of the hex vs. decimal mess with the frame counter.

Oh, I see.

So what about:
 if (bytes_read > count)
return -E-something?

Quoted 8 lines
Quoted 8 lines

 }

 static const struct file_operations i915_pipe_crc_fops = { --
2.5.5

Quoted 4 lines
Quoted 4 lines

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

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

#4 lore raw

Re: [PATCH 2/2] drm/i915: Fix the return value of pipe crc read function.

Quoted 40 lines · Rodrigo
Quoted 32 lines · Ville Syrjälä
Quoted 23 lines · Rodrigo Vivi

A read(fd, buf, len) function should return the number of bytes read. In our case we need to return the number of bytes we copy to user, instead of returning the number of bytes we read internally.

It was really strange when I saw i-g-t test case using len '54' but getting '56' as return. First thought was how do we read more than we asked? But also I checked and there was really only 54. Until I realized it was all our fault. EFAULT!

---
 drivers/gpu/drm/i915/i915_debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
index 7052c47..b7b8d79 100644

drivers/gpu/drm/i915/i915_debugfs.c i915_pipe_crc_read(struct file *filep, char · line 3663

Quoted 6 lines
drivers/gpu/drm/i915/i915_debugfs.c +1 -1
@@ -3659,7 +3659,7 @@ i915_pipe_crc_read(struct file *filep, char
 __user *user_buf, size_t count,
 
 	spin_unlock_irq(&pipe_crc->lock);
 
-	return bytes_read;
+	return PIPE_CRC_LINE_LEN;

Nope. We can read multiple entries in one go.

Quoted 6 lines

bytes_read += snprintf()

so it's mostly good. The only case where it fails if we don't have enough space for the snprintf(), as snprintf() will then return the number of bytes it would have written. That can actually happen on account of the hex vs. decimal mess with the frame counter.

Quoted 5 lines

Oh, I see.

So what about:
if (bytes_read > count)
return -E-something?

We could just return from snprintf() when it does not return PIPE_CRC_LINE_LEN instead of reading all entries.

Anyway, shouldn't we fix hex to decimal conv.? Or is the value incorrect when entry->frame in decimal exceeds 8 chars?

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
 }
 

static const struct file_operations i915_pipe_crc_fops = {

Signature · 1 line
2.5.5
Quoted 8 lines
Quoted 4 lines
Quoted 4 lines

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

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

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

#5 lore raw

Re: [PATCH 2/2] drm/i915: Fix the return value of pipe crc read function.

In reply to Rodrigo Vivi
Quoted 21 lines · Rodrigo Vivi

A read(fd, buf, len) function should return the number of bytes read. In our case we need to return the number of bytes we copy to user, instead of returning the number of bytes we read internally.

It was really strange when I saw i-g-t test case using len '54' but getting '56' as return. First thought was how do we read more than we asked? But also I checked and there was really only 54. Until I realized it was all our fault. EFAULT!

---
 drivers/gpu/drm/i915/i915_debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
index 7052c47..b7b8d79 100644

drivers/gpu/drm/i915/i915_debugfs.c i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count, · line 3662

Quoted 5 lines
drivers/gpu/drm/i915/i915_debugfs.c +1 -1
@@ -3659,7 +3659,7 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 
 	spin_unlock_irq(&pipe_crc->lock);
 
-	return bytes_read;
+	return PIPE_CRC_LINE_LEN;

This will no loose entries if you have more than 1 crc ready to be read. I think better to WARN_ON if snprintf doesn't give us the expected length, and then correct bytes_read to match.
-Daniel

Quoted 10 lines
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
 }
 
 static const struct file_operations i915_pipe_crc_fops = {
Signature · 6 lines
2.5.5

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#6 lore raw

✗ Ro.CI.BAT: failure for series starting with [1/2] drm/i915: Fix copy_to_user usage for pipe_crc

In reply to Rodrigo Vivi
== Series Details ==

== Series Details ==

Series: series starting with [1/2] drm/i915: Fix copy_to_user usage for pipe_crc
URL : https://patchwork.freedesktop.org/series/10573/
State : failure

== Summary ==

Series 10573v1 Series without cover letter
http://patchwork.freedesktop.org/api/1.0/series/10573/revisions/1/mbox

Test kms_cursor_legacy:

        Subgroup basic-cursor-vs-flip-legacy:
                pass       -> FAIL       (ro-byt-n2820)
        Subgroup basic-flip-vs-cursor-legacy:
                pass       -> FAIL       (fi-skl-i7-6700k)
Test kms_frontbuffer_tracking:
        Subgroup basic:
                pass       -> DMESG-FAIL (ro-bdw-i7-5557U)
                skip       -> FAIL       (ro-ilk1-i5-650)
                pass       -> DMESG-FAIL (ro-hsw-i3-4010u)
                pass       -> FAIL       (fi-skl-i5-6260u)
                pass       -> DMESG-FAIL (ro-bdw-i7-5600u)
                pass       -> FAIL       (ro-snb-i7-2620M)
                pass       -> DMESG-FAIL (ro-bdw-i5-5250u)
                pass       -> FAIL       (ro-hsw-i7-4770r)
                pass       -> FAIL       (fi-hsw-i7-4770k)
                pass       -> FAIL       (fi-snb-i7-2600)
                pass       -> FAIL       (ro-ivb-i7-3770)
                pass       -> FAIL       (ro-ivb2-i7-3770)
                pass       -> FAIL       (ro-byt-n2820)
                pass       -> FAIL       (fi-skl-i7-6700k)
                pass       -> FAIL       (ro-skl3-i5-6260u)
Test kms_pipe_crc_basic:
        Subgroup hang-read-crc-pipe-a:
                pass       -> FAIL       (ro-bdw-i7-5557U)
                pass       -> FAIL       (ro-ilk1-i5-650)
                pass       -> FAIL       (ro-hsw-i3-4010u)
                pass       -> FAIL       (fi-skl-i5-6260u)
                pass       -> FAIL       (ro-bdw-i7-5600u)
                pass       -> FAIL       (ro-snb-i7-2620M)
                pass       -> FAIL       (ro-bdw-i5-5250u)
                pass       -> FAIL       (ro-hsw-i7-4770r)
                pass       -> FAIL       (fi-snb-i7-2600)
                pass       -> FAIL       (ro-ivb-i7-3770)
                pass       -> FAIL       (ro-ivb2-i7-3770)
                pass       -> FAIL       (ro-byt-n2820)
                pass       -> FAIL       (ro-skl3-i5-6260u)
        Subgroup hang-read-crc-pipe-b:
                pass       -> FAIL       (ro-bdw-i7-5557U)
                pass       -> FAIL       (ro-ilk1-i5-650)
                pass       -> FAIL       (ro-hsw-i3-4010u)
                pass       -> FAIL       (fi-skl-i5-6260u)
                pass       -> FAIL       (ro-bdw-i7-5600u)
                pass       -> FAIL       (ro-snb-i7-2620M)
                pass       -> FAIL       (ro-bdw-i5-5250u)
                pass       -> FAIL       (ro-hsw-i7-4770r)
                pass       -> FAIL       (fi-snb-i7-2600)
                pass       -> FAIL       (ro-ivb-i7-3770)
                pass       -> FAIL       (ro-ivb2-i7-3770)
                pass       -> FAIL       (ro-byt-n2820)
                pass       -> FAIL       (ro-skl3-i5-6260u)
        Subgroup hang-read-crc-pipe-c:
                pass       -> FAIL       (ro-bdw-i7-5557U)
                pass       -> FAIL       (ro-hsw-i3-4010u)
                pass       -> FAIL       (fi-skl-i5-6260u)
                pass       -> FAIL       (ro-bdw-i7-5600u)
                pass       -> FAIL       (ro-bdw-i5-5250u)
                pass       -> FAIL       (ro-hsw-i7-4770r)
                pass       -> FAIL       (ro-ivb-i7-3770)
                pass       -> FAIL       (ro-ivb2-i7-3770)
                pass       -> FAIL       (ro-skl3-i5-6260u)
        Subgroup nonblocking-crc-pipe-a:
                pass       -> FAIL       (ro-bdw-i7-5557U)
                pass       -> FAIL       (ro-ilk1-i5-650)
                pass       -> FAIL       (ro-hsw-i3-4010u)
                pass       -> FAIL       (fi-skl-i5-6260u)
                pass       -> FAIL       (ro-bdw-i7-5600u)
                pass       -> FAIL       (ro-snb-i7-2620M)
                pass       -> FAIL       (ro-bdw-i5-5250u)
                pass       -> FAIL       (ro-hsw-i7-4770r)
                pass       -> FAIL       (fi-hsw-i7-4770k)
                pass       -> FAIL       (fi-snb-i7-2600)
                pass       -> FAIL       (ro-ivb-i7-3770)
                pass       -> FAIL       (ro-ivb2-i7-3770)
                pass       -> FAIL       (ro-byt-n2820)
                pass       -> FAIL       (fi-skl-i7-6700k)
                pass       -> FAIL       (ro-skl3-i5-6260u)
        Subgroup nonblocking-crc-pipe-a-frame-sequence:
                pass       -> FAIL       (ro-bdw-i7-5557U)
                pass       -> FAIL       (ro-ilk1-i5-650)
                pass       -> FAIL       (ro-hsw-i3-4010u)
                pass       -> FAIL       (fi-skl-i5-6260u)
                pass       -> FAIL       (ro-bdw-i7-5600u)
                pass       -> FAIL       (ro-snb-i7-2620M)
                pass       -> FAIL       (ro-bdw-i5-5250u)
                pass       -> FAIL       (ro-hsw-i7-4770r)
                pass       -> FAIL       (fi-hsw-i7-4770k)
                pass       -> FAIL       (fi-snb-i7-2600)
                pass       -> FAIL       (ro-ivb-i7-3770)
                pass       -> FAIL       (ro-ivb2-i7-3770)
                pass       -> FAIL       (ro-byt-n2820)
                pass       -> FAIL       (fi-skl-i7-6700k)
                pass       -> FAIL       (ro-skl3-i5-6260u)
        Subgroup nonblocking-crc-pipe-b:
                pass       -> FAIL       (ro-bdw-i7-5557U)
                pass       -> FAIL       (ro-ilk1-i5-650)
                pass       -> FAIL       (ro-hsw-i3-4010u)

WARNING: Long output truncated

Results at /archive/results/CI_IGT_test/RO_Patchwork_1673/

aa8628c drm-intel-nightly: 2016y-08m-02d-14h-10m-12s UTC integration manifest be0abf9 drm/i915: Fix the return value of pipe crc read function. e9fa366 drm/i915: Fix copy_to_user usage for pipe_crc

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

#7 lore raw
In reply to Rodrigo Vivi
Quoted 23 lines · Rodrigo Vivi

Copy to user return the number of bytes it couldn't write and zero on success. So any number different than 0 should be considered a fault, not only when it doesn't write the full size.

---
 drivers/gpu/drm/i915/i915_debugfs.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +0 -1
index 7c42ec4..7052c47 100644
@@ -3632,7 +3632,6 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 	while (n_entries > 0) {
 		struct intel_pipe_crc_entry *entry =
 			&pipe_crc->entries[pipe_crc->tail];
-		int ret;
 
 		if (CIRC_CNT(pipe_crc->head, pipe_crc->tail,
 			     INTEL_PIPE_CRC_ENTRIES_NR) < 1)

drivers/gpu/drm/i915/i915_debugfs.c i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count, · line 3651

Quoted 6 lines
drivers/gpu/drm/i915/i915_debugfs.c +1 -2
@@ -3649,8 +3648,7 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 
 		spin_unlock_irq(&pipe_crc->lock);
 
-		ret = copy_to_user(user_buf, buf, PIPE_CRC_LINE_LEN);
-		if (ret == PIPE_CRC_LINE_LEN)
+		if (!copy_to_user(user_buf, buf, PIPE_CRC_LINE_LEN))

A bad '!' snuck in.

Quoted 10 lines
drivers/gpu/drm/i915/i915_debugfs.c +0 -0
 			return -EFAULT;
 
 		user_buf += PIPE_CRC_LINE_LEN;
Signature · 6 lines
2.5.5

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/2
#8 lore raw
In reply to Rodrigo Vivi
Quoted 6 lines · Rodrigo Vivi

Copy to user return the number of bytes it couldn't write and zero on success. So any number different than 0 should be considered a fault, not only when it doesn't write the full size.

I audited all the other places we use copy_*_user. We have a bit a mess between negative errno and unwritten bytes for our own copy functions, but seems consistent (except this one here).

Quoted 33 lines
---
 drivers/gpu/drm/i915/i915_debugfs.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +1 -3
index 7c42ec4..7052c47 100644
@@ -3632,7 +3632,6 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 	while (n_entries > 0) {
 		struct intel_pipe_crc_entry *entry =
 			&pipe_crc->entries[pipe_crc->tail];
-		int ret;
 
 		if (CIRC_CNT(pipe_crc->head, pipe_crc->tail,
 			     INTEL_PIPE_CRC_ENTRIES_NR) < 1)
@@ -3649,8 +3648,7 @@ i915_pipe_crc_read(struct file *filep, char __user *user_buf, size_t count,
 
 		spin_unlock_irq(&pipe_crc->lock);
 
-		ret = copy_to_user(user_buf, buf, PIPE_CRC_LINE_LEN);
-		if (ret == PIPE_CRC_LINE_LEN)
+		if (!copy_to_user(user_buf, buf, PIPE_CRC_LINE_LEN))
 			return -EFAULT;
 
 		user_buf += PIPE_CRC_LINE_LEN;
Signature · 6 lines
2.5.5

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx