Skip to contentSkip to conversation

drm/i915/gen8: page directories rework allocation

PATCH Superseded Dan Carpenter 7 participants 13 messages Apr 30, 2015 – May 6, 2015

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi Michel,

The patch 69876bed7e00: "drm/i915/gen8: page directories rework allocation" from Apr 8, 2015, has the following issue:

	drivers/gpu/drm/i915/i915_gem_gtt.c:760
	warn: too many zeroes

drivers/gpu/drm/i915/i915_gem_gtt.c
746 static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,

   747                                       struct i915_page_directory_pointer *pdp,
   748                                       uint64_t start,
   749                                       uint64_t length,
   750                                       unsigned long *new_pds)
   751  {
   752          struct drm_device *dev = ppgtt->base.dev;
   753          struct i915_page_directory *pd;
   754          uint64_t temp;
   755          uint32_t pdpe;
   756  
   757          WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
   758  
   759          /* FIXME: PPGTT container_of won't work for 64b */
   760          WARN_ON((start + length) > 0x800000000ULL);
                                                     ^

This last zero was not intended.

   761  
   762          gen8_for_each_pdpe(pd, pdp, start, length, temp, pdpe) {
   763                  if (pd)
   764                          continue;
   765  
   766                  pd = alloc_pd_single(dev);
   767                  if (IS_ERR(pd))
   768                          goto unwind_out;
   769  
   770                  gen8_initialize_pd(&ppgtt->base, pd);
   771                  pdp->page_directory[pdpe] = pd;
   772                  set_bit(pdpe, new_pds);
   773          }
   774  
   775          return 0;
   776  
   777  unwind_out:
   778          for_each_set_bit(pdpe, new_pds, GEN8_LEGACY_PDPES)
   779                  unmap_and_free_pd(pdp->page_directory[pdpe], dev);
   780  
   781          return -ENOMEM;
   782  }

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

#1 lore raw
Quoted 25 lines · Dan Carpenter

Hi Michel,

The patch 69876bed7e00: "drm/i915/gen8: page directories rework allocation" from Apr 8, 2015, has the following issue:

	drivers/gpu/drm/i915/i915_gem_gtt.c:760
	warn: too many zeroes

drivers/gpu/drm/i915/i915_gem_gtt.c
746 static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,

   747                                       struct i915_page_directory_pointer *pdp,
   748                                       uint64_t start,
   749                                       uint64_t length,
   750                                       unsigned long *new_pds)
   751  {
   752          struct drm_device *dev = ppgtt->base.dev;
   753          struct i915_page_directory *pd;
   754          uint64_t temp;
   755          uint32_t pdpe;
   756  
   757          WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
   758  
   759          /* FIXME: PPGTT container_of won't work for 64b */
   760          WARN_ON((start + length) > 0x800000000ULL);
                                                     ^
Quoted 1 line

This last zero was not intended.

Wouldn't (1ULL << 31) be clearer?

And shouldn't it be >= not > ?

Or perhaps,

	/* FIXME: upper bound must not overflow 31 bits */
	WARN_ON((start + length) & (~0ULL << 31));

Pick whichever you think is most comprehensible and least typo-prone!

.Dave.

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

#2 lore raw

PATCH 1/1 drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

In reply to Dan Carpenter

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
index 6fae6bd..6d894fc 100644
@@ -756,8 +756,8 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
 	WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
-	/* FIXME: PPGTT container_of won't work for 64b */
-	WARN_ON((start + length) > 0x800000000ULL);
+	/* FIXME: upper bound must not overflow 31 bits  */
+	WARN_ON((start + length) & (~0ULL << 31));
 
 	gen8_for_each_pdpe(pd, pdp, start, length, temp, pdpe) {
 		if (pd)
Signature · 6 lines
2.1.1

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

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

Quoted 19 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6fae6bd..6d894fc 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt, · line 760

Quoted 7 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
@@ -756,8 +756,8 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
 	WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
-	/* FIXME: PPGTT container_of won't work for 64b */
-	WARN_ON((start + length) > 0x800000000ULL);
+	/* FIXME: upper bound must not overflow 31 bits  */
+	WARN_ON((start + length) & (~0ULL << 31));

Why is it 31 and not 32?

Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
 	gen8_for_each_pdpe(pd, pdp, start, length, temp, pdpe) {
 		if (pd)
Signature · 6 lines
2.1.1

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

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

Quoted 21 lines · Ville Syrjälä
Quoted 20 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

---
  drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6fae6bd..6d894fc 100644
@@ -756,8 +756,8 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
Quoted 9 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt, · line 760

Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
  	WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
-	/* FIXME: PPGTT container_of won't work for 64b */
-	WARN_ON((start + length) > 0x800000000ULL);
+	/* FIXME: upper bound must not overflow 31 bits  */
+	WARN_ON((start + length) & (~0ULL << 31));

Why is it 31 and not 32?

Right, the check really should be (start + length) >= 0x100000000ULL.

Quoted 11 lines
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -1
  	gen8_for_each_pdpe(pd, pdp, start, length, temp, pdpe) {
  		if (pd)
--

2.1.1

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

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

#5 lore raw

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

Quoted 26 lines · Michel Thierry
Quoted 25 lines · Ville Syrjälä
Quoted 24 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

---
  drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6fae6bd..6d894fc 100644
@@ -756,8 +756,8 @@ static int
 gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
      WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
Quoted 8 lines
Quoted 6 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt, · line 760

Quoted 4 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
-    /* FIXME: PPGTT container_of won't work for 64b */
-    WARN_ON((start + length) > 0x800000000ULL);
+    /* FIXME: upper bound must not overflow 31 bits  */
+    WARN_ON((start + length) & (~0ULL << 31));

Why is it 31 and not 32?

Right, the check really should be (start + length) >= 0x100000000ULL.

Something with '32' in it might be more obvious and save anyone having to count the zeroes ... and a comment that also mentioned the limit:

	/* FIXME: for now, upper bound must fit in 32 bits  */

	WARN_ON((start + length) >= (1ULL << 32))
	WARN_ON((start + length) & (~0ULL << 32))
	WARN_ON((start + length) >> 32) != 0)

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

#6 lore raw

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

Quoted 36 lines · Dave Gordon
Quoted 26 lines · Michel Thierry
Quoted 25 lines · Ville Syrjälä
Quoted 24 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

---
   drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
   1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6fae6bd..6d894fc 100644
@@ -756,8 +756,8 @@ static int
 gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
       WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
Quoted 8 lines
Quoted 6 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt, · line 760

Quoted 4 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
-    /* FIXME: PPGTT container_of won't work for 64b */
-    WARN_ON((start + length) > 0x800000000ULL);
+    /* FIXME: upper bound must not overflow 31 bits  */
+    WARN_ON((start + length) & (~0ULL << 31));

Why is it 31 and not 32?

Right, the check really should be (start + length) >= 0x100000000ULL.

Quoted 8 lines

Something with '32' in it might be more obvious and save anyone having to count the zeroes ... and a comment that also mentioned the limit:

	/* FIXME: for now, upper bound must fit in 32 bits  */

	WARN_ON((start + length) >= (1ULL << 32))
	WARN_ON((start + length) & (~0ULL << 32))
	WARN_ON((start + length) >> 32) != 0)

Yes, I was planning to just replace _31_ with _32_ and avoid all those zeroes...

Quoted 3 lines

.Dave.

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

#7 lore raw

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

In reply to Dave Gordon
Quoted 34 lines · Dave Gordon
Quoted 33 lines · Michel Thierry
Quoted 31 lines · Ville Syrjälä
Quoted 20 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

---
  drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6fae6bd..6d894fc 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static int · line 761

Quoted 8 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
@@ -756,8 +756,8 @@ static int
 gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
      WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
-    /* FIXME: PPGTT container_of won't work for 64b */
-    WARN_ON((start + length) > 0x800000000ULL);
+    /* FIXME: upper bound must not overflow 31 bits  */
+    WARN_ON((start + length) & (~0ULL << 31));

Why is it 31 and not 32?

Quoted 8 lines
Quoted 1 line

Right, the check really should be (start + length) >= 0x100000000ULL.

Something with '32' in it might be more obvious and save anyone having to count the zeroes ... and a comment that also mentioned the limit:

	/* FIXME: for now, upper bound must fit in 32 bits  */

	WARN_ON((start + length) >= (1ULL << 32))

This would match the '1ULL<<32' used in the ppgtt init.

Quoted 4 lines
	WARN_ON((start + length) & (~0ULL << 32))
	WARN_ON((start + length) >> 32) != 0)

.Dave.

Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Tested-by on v1 1/1
#8 lore raw

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

In reply to Michel Thierry

Task id: 6299 -------------------------------------Summary-------------------------------------
Platform Delta drm-intel-nightly Series Applied

PNV                                  276/276              276/276
ILK                                  302/302              302/302
SNB                                  316/316              316/316
IVB                                  264/264              264/264

BYT -3 227/227 224/227
BDW 318/318 318/318
-------------------------------------Detailed-------------------------------------
Platform Test drm-intel-nightly Series Applied
*BYT igt@gem_dummy_reloc_loop@render FAIL(1)PASS(18) TIMEOUT(1)PASS(1)
*BYT igt@gem_exec_parse@bitmasks FAIL(1)PASS(7) DMESG_WARN(1)PASS(1)
(dmesg patch applied)drm:check_crtc_state[i915]]*ERROR*mismatch_in_has_infoframe(expected#,found#)@mismatch in has_infoframe .* found WARNING:at_drivers/gpu/drm/i915/intel_display.c:#check_crtc_state[i915]()@WARNING:.* at .* check_crtc_state+0x
BYT igt@gem_pipe_control_store_loop@fresh-buffer FAIL(1)TIMEOUT(10)PASS(9) TIMEOUT(2)
Note: You need to pay more attention to line start with '*'
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#9 lore raw

Re: [PATCH] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

In reply to Michel Thierry
Quoted 3 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead.

I think you mean "less typo-prone" here :)

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

#10 lore raw

PATCH v2 1/1 drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

In reply to Dan Carpenter

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead, to check for (start + length) >= 0x100000000ULL.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

v2: Really check for 32b overflow (Ville)

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
index 6fae6bd..3450cb5 100644
@@ -756,8 +756,8 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
 	WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
-	/* FIXME: PPGTT container_of won't work for 64b */
-	WARN_ON((start + length) > 0x800000000ULL);
+	/* FIXME: upper bound must not overflow 32 bits  */
+	WARN_ON((start + length) >= (1ULL << 32));
 
 	gen8_for_each_pdpe(pd, pdp, start, length, temp, pdpe) {
 		if (pd)
Signature · 6 lines
2.1.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Tested-by on v2 1/1
#11 lore raw

Re: [PATCH v2] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

Task id: 6300 -------------------------------------Summary-------------------------------------
Platform Delta drm-intel-nightly Series Applied

PNV                                  276/276              276/276
ILK                                  302/302              302/302
SNB                                  316/316              316/316
IVB                                  264/264              264/264

BYT -3 227/227 224/227
BDW 318/318 318/318
-------------------------------------Detailed-------------------------------------
Platform Test drm-intel-nightly Series Applied

*BYT  igt@gem_dummy_reloc_loop@render      FAIL(1)PASS(18)      TIMEOUT(1)PASS(1)
 BYT  igt@gem_pipe_control_store_loop@fresh-buffer      FAIL(1)TIMEOUT(10)PASS(9)      TIMEOUT(1)PASS(1)
*BYT  igt@gem_threaded_access_tiled      FAIL(1)PASS(6)      DMESG_WARN(1)PASS(1)

(dmesg patch applied)drm:check_crtc_state[i915]]*ERROR*mismatch_in_has_infoframe(expected#,found#)@mismatch in has_infoframe .* found WARNING:at_drivers/gpu/drm/i915/intel_display.c:#check_crtc_state[i915]()@WARNING:.* at .* check_crtc_state+0x
Note: You need to pay more attention to line start with '*'
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#12 lore raw

Re: [PATCH v2] drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories

In reply to Michel Thierry
Quoted 3 lines · Michel Thierry

The patch 69876bed7e008f5fe01538a2d47c09f2862129d0: "drm/i915/gen8:
page directories rework allocation" added an overflow warning, but the mask had an extra 0. Use typo-prone option suggested by Dave instead,

I inserted the "less" above while applying.

Quoted 11 lines

to check for (start + length) >= 0x100000000ULL.

This check will be unnecessary after gen8_alloc_va_range handles more than 4 PDPs (48b addressing).

v2: Really check for 32b overflow (Ville)

Queued for -next, thanks for the patch.
-Daniel

Quoted 26 lines
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
index 6fae6bd..3450cb5 100644
@@ -756,8 +756,8 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,
 
 	WARN_ON(!bitmap_empty(new_pds, GEN8_LEGACY_PDPES));
 
-	/* FIXME: PPGTT container_of won't work for 64b */
-	WARN_ON((start + length) > 0x800000000ULL);
+	/* FIXME: upper bound must not overflow 32 bits  */
+	WARN_ON((start + length) >= (1ULL << 32));
 
 	gen8_for_each_pdpe(pd, pdp, start, length, temp, pdpe) {
 		if (pd)
Signature · 6 lines
2.1.1

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