Skip to contentSkip to conversation

drm/radeon: add radeon_atom_get_clock_dividers helper

PATCH In review Dan Carpenter 4 participants 10 messages Apr 18, 2013 – Apr 22, 2013

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

Hello Christian König,

The patch 7062ab67d4c6: "drm/radeon: add radeon_atom_get_clock_dividers helper" from Apr 8, 2013, has endian bugs.

drivers/gpu/drm/radeon/radeon_atombios.c
2712 if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
2713 args.v3.ulClock.ulComputeClockFlag = clock_type;
2714 args.v3.ulClock.ulClockFreq = cpu_to_le32(clock); /* 10 khz */

                                                        ^^^^^^^^^^^

This is 24 bit bitfield so it can't store a __le32. On little endian systems it will truncate high bits away so that's ok, but on big endian it will break.

  2715  
  2716                                  atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
  2717  
  2718                                  dividers->post_div = args.v3.ucPostDiv;
  2719                                  dividers->enable_post_div = (args.v3.ucCntlFlag &
  2720                                                               ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;

There are a lot of other Sparse endian warnings but I haven't looked at them. Here is how to check: http://lwn.net/Articles/205624/

regards,
dan carpenter

#1 lore raw

Am 18.04.2013 20:47, schrieb Dan Carpenter:

Quoted 22 lines

Hello Christian König,

The patch 7062ab67d4c6: "drm/radeon: add radeon_atom_get_clock_dividers helper" from Apr 8, 2013, has endian bugs.

drivers/gpu/drm/radeon/radeon_atombios.c

   2712                          if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
   2713                                  args.v3.ulClock.ulComputeClockFlag = clock_type;
   2714                                  args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);       /* 10 khz */
                                                         ^^^^^^^^^^^

This is 24 bit bitfield so it can't store a __le32. On little endian systems it will truncate high bits away so that's ok, but on big endian it will break.

   2715
   2716                                  atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
   2717
   2718                                  dividers->post_div = args.v3.ucPostDiv;
   2719                                  dividers->enable_post_div = (args.v3.ucCntlFlag &
   2720                                                               ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
Quoted 2 lines

There are a lot of other Sparse endian warnings but I haven't looked at them. Here is how to check: http://lwn.net/Articles/205624/

Hi Dan,

in general I'm not sure that we support any big endian system with that code any more.

But Alex wrote most of the atombios code, so that is probably more a question for him.

Christian.

Quoted 4 lines

regards,
dan carpenter

#2 lore raw

PATCH 1/1 drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

In reply to Dan Carpenter

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +2 -0
index 4b04ba3..de678dd 100644
@@ -459,6 +459,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
   UCHAR   ucPostDiv;                          //Output Parameter      
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
   UCHAR   ucPostDiv;                          //Output Parameter      
drivers/gpu/drm/radeon/radeon_atombios.c +2 -4
index 8c1779c..2fc444e 100644
@@ -2710,8 +2710,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->enable_post_div = (dividers->fb_div & 1) ? true : false;
 		} else {
 			if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
-				args.v3.ulClock.ulComputeClockFlag = clock_type;
-				args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);
 
 				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 
@@ -2726,8 +2725,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->vco_mode = (args.v3.ucCntlFlag &
 						      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
 			} else {
-				args.v5.ulClock.ulComputeClockFlag = clock_type;
-				args.v5.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);
 				if (strobe_mode)
 					args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
 
Signature · 1 line
1.7.7.5
#3 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

Quoted 14 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +0 -0
index 4b04ba3..de678dd 100644

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 · line 462

Quoted 4 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
@@ -459,6 +459,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;

Why is this a long instead of an __le32 or u32?

Quoted 26 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
   UCHAR   ucPostDiv;                          //Output Parameter      
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
   UCHAR   ucPostDiv;                          //Output Parameter      
drivers/gpu/drm/radeon/radeon_atombios.c +1 -2
index 8c1779c..2fc444e 100644
@@ -2710,8 +2710,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->enable_post_div = (dividers->fb_div & 1) ? true : false;
 		} else {
 			if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
-				args.v3.ulClock.ulComputeClockFlag = clock_type;
-				args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);
 
 				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 

drivers/gpu/drm/radeon/radeon_atombios.c int radeon_atom_get_clock_dividers(struct radeon_device *rdev, · line 2728

Quoted 6 lines
drivers/gpu/drm/radeon/radeon_atombios.c +1 -2
@@ -2726,8 +2725,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->vco_mode = (args.v3.ucCntlFlag &
 						      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
 			} else {
-				args.v5.ulClock.ulComputeClockFlag = clock_type;
-				args.v5.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);

We've changed from v5 to v3. Was that intentional?

I'm confused by this patch as well. I assumed the datatypes were determined by the hardware spec.

regards,
dan carpenter

#4 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

Quoted 10 lines · Dan Carpenter
Quoted 9 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
Quoted 11 lines

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 · line 462

Quoted 9 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
index 4b04ba3..de678dd 100644
@@ -459,6 +459,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;

Why is this a long instead of an __le32 or u32?

atombios.h is shared across OSes and has it's own types.

Quoted 27 lines
Quoted 27 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
   };
   UCHAR   ucRefDiv;                           //Output Parameter
   UCHAR   ucPostDiv;                          //Output Parameter
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter
   UCHAR   ucPostDiv;                          //Output Parameter
drivers/gpu/drm/radeon/radeon_atombios.c +1 -2
index 8c1779c..2fc444e 100644
@@ -2710,8 +2710,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
                              dividers->enable_post_div = (dividers->fb_div & 1) ? true : false;
              } else {
                      if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
-                             args.v3.ulClock.ulComputeClockFlag = clock_type;
-                             args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);       /* 10 khz */
+                             args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);
 
                              atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 
Quoted 9 lines

drivers/gpu/drm/radeon/radeon_atombios.c int radeon_atom_get_clock_dividers(struct radeon_device *rdev, · line 2728

Quoted 7 lines
drivers/gpu/drm/radeon/radeon_atombios.c +1 -2
@@ -2726,8 +2725,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
                              dividers->vco_mode = (args.v3.ucCntlFlag &
                                                    ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
                      } else {
-                             args.v5.ulClock.ulComputeClockFlag = clock_type;
-                             args.v5.ulClock.ulClockFreq = cpu_to_le32(clock);       /* 10 khz */
+                             args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);

We've changed from v5 to v3. Was that intentional?

Copy paste typo. fixed in v2 which is attached.

Quoted 3 lines

I'm confused by this patch as well. I assumed the datatypes were determined by the hardware spec.

I'm not sure I follow. The atombios interpretor requires data in little endian format.

Alex

From ce94ed1083df11895dbec520ef243d4fd805a4a9 Mon Sep 17 00:00:00 2001
From: Alex Deucher <[email protected]>
Date: Mon, 22 Apr 2013 09:59:01 -0400
Subject: [PATCH] drm/radeon: fix endian bugs in
radeon_atom_get_clock_dividers() (v2)

v2: fix copy paste typo.

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +2 -0
index 4b04ba3..de678dd 100644
@@ -459,6 +459,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
   UCHAR   ucPostDiv;                          //Output Parameter      
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
   UCHAR   ucPostDiv;                          //Output Parameter      
drivers/gpu/drm/radeon/radeon_atombios.c +2 -4
index 8c1779c..4b853d8 100644
@@ -2710,8 +2710,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->enable_post_div = (dividers->fb_div & 1) ? true : false;
 		} else {
 			if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
-				args.v3.ulClock.ulComputeClockFlag = clock_type;
-				args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v3.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);
 
 				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 
@@ -2726,8 +2725,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->vco_mode = (args.v3.ucCntlFlag &
 						      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
 			} else {
-				args.v5.ulClock.ulComputeClockFlag = clock_type;
-				args.v5.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v5.ulClockFbDiv = cpu_to_le32((clock_type << 24) | clock);
 				if (strobe_mode)
 					args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
 
Signature · 1 line
1.7.7.5

_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel

#5 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

Quoted 24 lines · Alex Deucher
Quoted 10 lines · Dan Carpenter
Quoted 9 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
Quoted 11 lines

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 · line 462

Quoted 9 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
index 4b04ba3..de678dd 100644
@@ -459,6 +459,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;

Why is this a long instead of an __le32 or u32?

Quoted 1 line

atombios.h is shared across OSes and has it's own types.

Can ULONG be 64 bit? It's ugly when ULONG, ulong and "unsigned long" are different types.

Quoted 6 lines
Quoted 3 lines

I'm confused by this patch as well. I assumed the datatypes were determined by the hardware spec.

I'm not sure I follow. The atombios interpretor requires data in little endian format.

I was expecting that the code would stay the same and the annotations would change is all... I haven't tested this so I'm sure it's right, but it just wasn't the change I was expecting.

regards,
dan carpenter

#6 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

Quoted 25 lines · Dan Carpenter
Quoted 24 lines · Alex Deucher
Quoted 21 lines · Dan Carpenter
Quoted 14 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +0 -0
index 4b04ba3..de678dd 100644

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 · line 462

Quoted 4 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
@@ -459,6 +459,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;

Why is this a long instead of an __le32 or u32?

Quoted 4 lines
Quoted 1 line

atombios.h is shared across OSes and has it's own types.

Can ULONG be 64 bit? It's ugly when ULONG, ulong and "unsigned long" are different types.

The atombios ULONG type is always 32 bits.

Quoted 11 lines
Quoted 7 lines
Quoted 4 lines

I'm confused by this patch as well. I assumed the datatypes were determined by the hardware spec.

I'm not sure I follow. The atombios interpretor requires data in little endian format.

I was expecting that the code would stay the same and the annotations would change is all... I haven't tested this so I'm sure it's right, but it just wasn't the change I was expecting.

This seemed like the most obvious solution to me, but it may not be optimal. What would you have done?

Alex

#7 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

In reply to Alex Deucher
Quoted 16 lines · Alex Deucher
Quoted 14 lines · Dan Carpenter
Quoted 13 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +0 -0
index 4b04ba3..de678dd 100644

[...]

Quoted 8 lines
Quoted 8 lines

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 · line 498

Quoted 8 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter
   UCHAR   ucPostDiv;                          //Output Parameter

This may just be a nitpick, but the location and name of the new union member is slightly confusing: It suggests that it's some kind of combination of the ulClock and ulFbDiv members, when it's just an alternative representation of ulClock. I'd suggest moving up the new member, clarifying its name and/or adding a comment explaining what it is for.
Looks good to me other than that.

Signature · 6 lines
Earthling Michel Dänzer           |                   http://www.amd.com
Libre software enthusiast         |          Debian, X and DRI developer
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
#8 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

Quoted 18 lines · Michel Dänzer
Quoted 16 lines · Alex Deucher
Quoted 14 lines · Dan Carpenter
Quoted 13 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +0 -0
index 4b04ba3..de678dd 100644

[...]

Quoted 16 lines
Quoted 8 lines
Quoted 8 lines

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 · line 498

Quoted 8 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter
   UCHAR   ucPostDiv;                          //Output Parameter

This may just be a nitpick, but the location and name of the new union member is slightly confusing: It suggests that it's some kind of combination of the ulClock and ulFbDiv members, when it's just an alternative representation of ulClock. I'd suggest moving up the new member, clarifying its name and/or adding a comment explaining what it is for.
Looks good to me other than that.

Sounds good. v3 attached.

Alex

From d21f9d7ef1645fcb1573c80a8b0160cd3208a22a Mon Sep 17 00:00:00 2001
From: Alex Deucher <[email protected]>
Date: Mon, 22 Apr 2013 09:59:01 -0400
Subject: [PATCH] drm/radeon: fix endian bugs in
radeon_atom_get_clock_dividers() (v3)

v2: fix copy paste typo.
v3: clarify new union member

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +2 -0
index 4b04ba3..0ee5737 100644
@@ -458,6 +458,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3
   union
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
+    ULONG ulClockParams;                      //ULONG access for BE
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
@@ -490,6 +491,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   union
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
+    ULONG ulClockParams;                      //ULONG access for BE
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
   };
   UCHAR   ucRefDiv;                           //Output Parameter      
drivers/gpu/drm/radeon/radeon_atombios.c +2 -4
index 8c1779c..0dd87c0 100644
@@ -2710,8 +2710,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->enable_post_div = (dividers->fb_div & 1) ? true : false;
 		} else {
 			if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
-				args.v3.ulClock.ulComputeClockFlag = clock_type;
-				args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v3.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
 
 				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 
@@ -2726,8 +2725,7 @@ int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
 				dividers->vco_mode = (args.v3.ucCntlFlag &
 						      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
 			} else {
-				args.v5.ulClock.ulComputeClockFlag = clock_type;
-				args.v5.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				args.v5.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
 				if (strobe_mode)
 					args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
 
Signature · 1 line
1.7.7.5

_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Reviewed-by
#9 lore raw

Re: [PATCH] drm/radeon: fix endian bugs in radeon_atom_get_clock_dividers()

Quoted 36 lines · Alex Deucher
Quoted 18 lines · Michel Dänzer
Quoted 16 lines · Alex Deucher
Quoted 14 lines · Dan Carpenter
Quoted 13 lines · On Mon, Apr 22, 2013 at 10:03:13AM -0400, [email protected] wrote:

From: Alex Deucher <[email protected]>

---
 drivers/gpu/drm/radeon/atombios.h        |    2 ++
 drivers/gpu/drm/radeon/radeon_atombios.c |    6 ++----
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/radeon/atombios.h +0 -0
index 4b04ba3..de678dd 100644

[...]

Quoted 16 lines
Quoted 8 lines
Quoted 8 lines

drivers/gpu/drm/radeon/atombios.h typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 · line 498

Quoted 8 lines
drivers/gpu/drm/radeon/atombios.h +1 -0
@@ -491,6 +492,7 @@ typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5
   {
     ATOM_COMPUTE_CLOCK_FREQ  ulClock;         //Input Parameter
     ATOM_S_MPLL_FB_DIVIDER   ulFbDiv;         //Output Parameter
+    ULONG ulClockFbDiv;
   };
   UCHAR   ucRefDiv;                           //Output Parameter
   UCHAR   ucPostDiv;                          //Output Parameter

This may just be a nitpick, but the location and name of the new union member is slightly confusing: It suggests that it's some kind of combination of the ulClock and ulFbDiv members, when it's just an alternative representation of ulClock. I'd suggest moving up the new member, clarifying its name and/or adding a comment explaining what it is for.
Looks good to me other than that.

Quoted 1 line

Sounds good. v3 attached.

Thanks.

Signature · 6 lines
Earthling Michel Dänzer           |                   http://www.amd.com
Libre software enthusiast         |          Debian, X and DRI developer
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel