ALSA: core: trivial code style fix
PATCH 1/3 ALSA: core: trivial code style fix
remove trailing tab on the line.
--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
sound/core/control.c +1 -1
@@ -1072,7 +1072,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,long private_size;struct user_element *ue;int idx, err; - +if (card->user_ctl_count >=MAX_USER_CONTROLS )return -ENOMEM ;if (info->count <1 )
PATCH 2/3 ALSA: core: release the constraint check for replace ops
Suppose the ALSA card already has a number of MAX_USER_CONTROLS controls, and the user wants to replace one, it should not fail at this condition check.
--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
sound/core/control.c +1 -1
@@ -1073,7 +1073,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,struct user_element *ue;int idx, err; -if (card->user_ctl_count >=MAX_USER_CONTROLS ) +if (card->user_ctl_count >=MAX_USER_CONTROLS + !!replace)return -ENOMEM ;if (info->count <1 )return -EINVAL ;
Re: [PATCH 2/3] ALSA: core: release the constraint check for replace ops
At Wed, 24 Aug 2011 11:12:39 +0800,
Quoted 3 lines · Lu Guanqun
Suppose the ALSA card already has a number of
MAX_USER_CONTROLScontrols, and the user wants to replace one, it should not fail at this condition check.
It'd be simpler to ignore the check when replace flag is set? The proposed change here looks too tricky, IMO...
thanks,
Takashi
Quoted 19 lines
Signed-off-by Lu Guanqun <[email protected]>--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)sound/core/control.c +1
-1@@ -1073,7 +1073,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,struct user_element *ue;int idx, err; -if (card->user_ctl_count >=MAX_USER_CONTROLS ) +if (card->user_ctl_count >=MAX_USER_CONTROLS + !!replace)return -ENOMEM ;if (info->count <1 )return -EINVAL ;
Re: [PATCH 2/3] ALSA: core: release the constraint check for replace ops
Quoted 8 lines · Takashi Iwai
At Wed, 24 Aug 2011 11:12:39 +0800,
Quoted 3 lines · Lu Guanqun
Suppose the ALSA card already has a number of
MAX_USER_CONTROLScontrols, and the user wants to replace one, it should not fail at this condition check.It'd be simpler to ignore the check when replace flag is set? The proposed change here looks too tricky, IMO...
Agreed. I'll cook another one.
Signature · 1 line
guanqun
PATCH v2 1/1 ALSA: core: release the constraint check for replace ops
Suppose the ALSA card already has a number of MAX_USER_CONTROLS controls, and the user wants to replace one, it should not fail at this condition check.
--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
sound/core/control.c +1 -1
@@ -1073,7 +1073,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,struct user_element *ue;int idx, err; -if (card->user_ctl_count >=MAX_USER_CONTROLS ) +if (!replace && card->user_ctl_count >=MAX_USER_CONTROLS )return -ENOMEM ;if (info->count <1 )return -EINVAL ;
Takashi Iwai
AppliedRe: [PATCH v2] ALSA: core: release the constraint check for replace ops
At Wed, 24 Aug 2011 14:45:10 +0800,
Quoted 5 lines · Lu Guanqun
Suppose the ALSA card already has a number of
MAX_USER_CONTROLScontrols, and the user wants to replace one, it should not fail at this condition check.Signed-off-by Lu Guanqun <[email protected]>
Thanks, applied.
Takashi
Quoted 18 lines
--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)sound/core/control.c +1
-1@@ -1073,7 +1073,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,struct user_element *ue;int idx, err; -if (card->user_ctl_count >=MAX_USER_CONTROLS ) +if (!replace && card->user_ctl_count >=MAX_USER_CONTROLS )return -ENOMEM ;if (info->count <1 )return -EINVAL ;
PATCH 3/3 ALSA: core: set kcontrol's count field correctly
In reply to Lu GuanqunI don't see how info's owner field relates to kcontrol's count field. It should assign to info's count instead.
Let's assume this scenario:
1. user reads the control element from kernel (the owner field is set)
2. user changes some values
3. user issues 'SNDRV_CTL_IOCTL_ELEM_REPLACE' ioctl.
With the original code, 'kctl' here gets a large count number due to its non-empty owner field. Therefore it fails on subsequent call snd_ctl_new().
This patch fixes this issue.
--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
sound/core/control.c +1 -1
@@ -1099,7 +1099,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,if (err <0 )return err;memcpy (&kctl.id, &info->id,sizeof (info->id)); - kctl.count = info->owner ? info->owner :1 ; + kctl.count = info->count; access |=SNDRV_CTL_ELEM_ACCESS_USER ; kctl.info = snd_ctl_elem_user_info;if (access &SNDRV_CTL_ELEM_ACCESS_READ )
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
At Wed, 24 Aug 2011 11:12:43 +0800,
Quoted 4 lines · Lu Guanqun
I don't see how info's owner field relates to kcontrol's count field. It should assign to info's count instead.
Quoted 8 lines
Let's assume this scenario:
1. user reads the control element from kernel (the owner field is set)
2. user changes some values
3. user issues 'SNDRV_CTL_IOCTL_ELEM_REPLACE' ioctl.With the original code, 'kctl' here gets a large count number due to its non-empty owner field. Therefore it fails on subsequent call
snd_ctl_new().
In IOCTL_ELEM_ADD and REPLACE, count and owner struct fields have different meanings from others. The count contains the array size, not the number of identical elements. And the owner contains the number of elements to be created. So, the current code is correct.
What we need is the documentation of this feature...
thanks,
Takashi
Quoted 21 lines
This patch fixes this issue.
Signed-off-by Lu Guanqun <[email protected]>--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)sound/core/control.c +1
-1@@ -1099,7 +1099,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,if (err <0 )return err;memcpy (&kctl.id, &info->id,sizeof (info->id)); - kctl.count = info->owner ? info->owner :1 ; + kctl.count = info->count; access |=SNDRV_CTL_ELEM_ACCESS_USER ; kctl.info = snd_ctl_elem_user_info;if (access &SNDRV_CTL_ELEM_ACCESS_READ )
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
Quoted 12 lines · Takashi Iwai
At Wed, 24 Aug 2011 11:12:43 +0800,
Quoted 10 lines · Lu Guanqun
I don't see how info's owner field relates to kcontrol's count field. It should assign to info's count instead.
Let's assume this scenario:
1. user reads the control element from kernel (the owner field is set)
2. user changes some values
3. user issues 'SNDRV_CTL_IOCTL_ELEM_REPLACE' ioctl.
Quoted 7 lines
Quoted 2 lines
With the original code, 'kctl' here gets a large count number due to its non-empty owner field. Therefore it fails on subsequent call
snd_ctl_new().In
IOCTL_ELEM_ADDand REPLACE, count and owner struct fields have different meanings from others. The count contains the array size, not the number of identical elements. And the owner contains the number of elements to be created. So, the current code is correct.
This sounds tricky...
in snd_ctl_elem_info():
info->owner = pid_vnr(vd->owner->pid);
} else {
info->owner = -1;This is where owner of info get assigned, but it's not number of elements to be created. I'm a bit unclear on this.
Quoted 2 lines
What we need is the documentation of this feature...
So my test case seems to be wrong, but how should I change it? Do I need to clear the owner field when user space application sends down the REPLACE ioctl?
Signature · 1 line
guanqun
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
At Wed, 24 Aug 2011 14:17:57 +0800,
Quoted 24 lines · Lu Guanqun
Quoted 12 lines · Takashi Iwai
At Wed, 24 Aug 2011 11:12:43 +0800,
Quoted 10 lines · Lu Guanqun
I don't see how info's owner field relates to kcontrol's count field. It should assign to info's count instead.
Let's assume this scenario:
1. user reads the control element from kernel (the owner field is set)
2. user changes some values
3. user issues 'SNDRV_CTL_IOCTL_ELEM_REPLACE' ioctl.Quoted 7 lines
Quoted 2 lines
With the original code, 'kctl' here gets a large count number due to its non-empty owner field. Therefore it fails on subsequent call
snd_ctl_new().In
IOCTL_ELEM_ADDand REPLACE, count and owner struct fields have different meanings from others. The count contains the array size, not the number of identical elements. And the owner contains the number of elements to be created. So, the current code is correct.This sounds tricky...
Quoted 7 lines
in
snd_ctl_elem_info():info->owner = pid_vnr(vd->owner->pid); } else { info->owner = -1;This is where owner of info get assigned, but it's not number of elements to be created.
The kcontrol.count field is never exposed to user-space.
Quoted 8 lines
I'm a bit unclear on this.
Quoted 2 lines
What we need is the documentation of this feature...
So my test case seems to be wrong, but how should I change it? Do I need to clear the owner field when user space application sends down the REPLACE ioctl?
Yes. Maybe it'd be better to define a new struct and use it in ADD/REPLACE ioctls so that user sees a clear difference. The struct will be identical except for renaming from owner to elem_count or such.
Takashi
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
At Wed, 24 Aug 2011 08:26:47 +0200,
Quoted 42 lines · Takashi Iwai
At Wed, 24 Aug 2011 14:17:57 +0800,
Quoted 24 lines · Lu Guanqun
Quoted 12 lines · Takashi Iwai
At Wed, 24 Aug 2011 11:12:43 +0800,
Quoted 10 lines · Lu Guanqun
I don't see how info's owner field relates to kcontrol's count field. It should assign to info's count instead.
Let's assume this scenario:
1. user reads the control element from kernel (the owner field is set)
2. user changes some values
3. user issues 'SNDRV_CTL_IOCTL_ELEM_REPLACE' ioctl.Quoted 7 lines
Quoted 2 lines
With the original code, 'kctl' here gets a large count number due to its non-empty owner field. Therefore it fails on subsequent call
snd_ctl_new().In
IOCTL_ELEM_ADDand REPLACE, count and owner struct fields have different meanings from others. The count contains the array size, not the number of identical elements. And the owner contains the number of elements to be created. So, the current code is correct.This sounds tricky...
Quoted 7 lines
in
snd_ctl_elem_info():info->owner = pid_vnr(vd->owner->pid); } else { info->owner = -1;This is where owner of info get assigned, but it's not number of elements to be created.
The kcontrol.count field is never exposed to user-space.
Quoted 5 lines
I'm a bit unclear on this.
Quoted 2 lines
What we need is the documentation of this feature...
Quoted 8 lines
Quoted 3 lines
So my test case seems to be wrong, but how should I change it? Do I need to clear the owner field when user space application sends down the REPLACE ioctl?
Yes. Maybe it'd be better to define a new struct and use it in ADD/REPLACE ioctls so that user sees a clear difference. The struct will be identical except for renaming from owner to
elem_countor such.
I mean a patch like below (untested).
Takashi
---
include/sound/asound.h +29 -2
@@ -813,6 +813,33 @@ struct snd_ctl_elem_info {unsigned char reserved[64 -4 *sizeof (unsigned short )]; }; +/* The struct used for ELEM_ADD and ELEM_REPLACE ioctls; +* This is is identical with snd_ctl_elem_info, but just owner field is +* different, keeping the number of elements to be created. +* The unused fields are stripped for simplicity. +*/ +struct snd_ctl_elem_add_info { +struct snd_ctl_elem_id id;/* element ID */ +snd_ctl_elem_type_t type;/* value type */ +unsigned int access;/* value access (bitmask) */ +unsigned int count;/* count of values */ +__kernel_pid_t elem_count;/* # of elements to create (0=1) */ +union { +struct { +long min;/* R: minimum value */ +long max;/* R: maximum value */ +long step;/* R: step (0 variable) */ + } integer; +struct { +long long min;/* R: minimum value */ +long long max;/* R: maximum value */ +long long step;/* R: step (0 variable) */ + } integer64; +unsigned char reserved[128 ]; + } value; +unsigned char reserved[64 ]; +}; +struct snd_ctl_elem_value {struct snd_ctl_elem_id id;/* W: element ID */ unsigned int indirect:1 ;/* W: indirect access - obsoleted */
@@ -854,8 +881,8 @@ struct snd_ctl_tlv {#define SNDRV_CTL_IOCTL_ELEM_LOCK _IOW('U', 0x14, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_ELEM_UNLOCK _IOW('U', 0x15, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_SUBSCRIBE_EVENTS _IOWR('U', 0x16, int) -#define SNDRV_CTL_IOCTL_ELEM_ADD _IOWR('U', 0x17, struct snd_ctl_elem_info) -#define SNDRV_CTL_IOCTL_ELEM_REPLACE _IOWR('U', 0x18, struct snd_ctl_elem_info) +#define SNDRV_CTL_IOCTL_ELEM_ADD _IOWR('U', 0x17, struct snd_ctl_elem_add_info) +#define SNDRV_CTL_IOCTL_ELEM_REPLACE _IOWR('U', 0x18, struct snd_ctl_elem_add_info)#define SNDRV_CTL_IOCTL_ELEM_REMOVE _IOWR('U', 0x19, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_TLV_READ _IOWR('U', 0x1a, struct snd_ctl_tlv)#define SNDRV_CTL_IOCTL_TLV_WRITE _IOWR('U', 0x1b, struct snd_ctl_tlv)
sound/core/control.c +6 -5
@@ -1064,7 +1064,7 @@ static void snd_ctl_elem_user_free(struct snd_kcontrol *kcontrol) }static int snd_ctl_elem_add (struct snd_ctl_file *file, -struct snd_ctl_elem_info *info,int replace) +struct snd_ctl_elem_add_info *info,int replace) {struct snd_card *card = file->card;struct snd_kcontrol kctl, *_kctl;
@@ -1099,7 +1099,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,if (err <0 )return err;memcpy (&kctl.id, &info->id,sizeof (info->id)); - kctl.count = info->owner ? info->owner :1 ; + kctl.count = info->elem_count ? info->elem_count :1 ; access |=SNDRV_CTL_ELEM_ACCESS_USER ; kctl.info = snd_ctl_elem_user_info;if (access &SNDRV_CTL_ELEM_ACCESS_READ )
@@ -1139,7 +1139,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file, ue =kzalloc (sizeof (struct user_element ) + private_size,GFP_KERNEL );if (ue ==NULL )return -ENOMEM ; - ue->info = *info; +memcpy (&ue->info, info,sizeof (*info)); ue->info.access =0 ; ue->elem_data = (char *)ue +sizeof (*ue); ue->elem_data_size = private_size;
@@ -1164,9 +1164,10 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file, }static int snd_ctl_elem_add_user (struct snd_ctl_file *file, -struct snd_ctl_elem_info __user *_info,int replace) +struct snd_ctl_elem_add_info __user *_info, +int replace) { -struct snd_ctl_elem_info info; +struct snd_ctl_elem_add_info info;if (copy_from_user (&info, _info,sizeof (info)))return -EFAULT ;return snd_ctl_elem_add (file, &info, replace);
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
Quoted 64 lines · Takashi Iwai
---
include/sound/asound.h +29
-2@@ -813,6 +813,33 @@ struct snd_ctl_elem_info {unsigned char reserved[64 -4 *sizeof (unsigned short )]; }; +/* The struct used for ELEM_ADD and ELEM_REPLACE ioctls; +* This is is identical with snd_ctl_elem_info, but just owner field is +* different, keeping the number of elements to be created. +* The unused fields are stripped for simplicity. +*/ +struct snd_ctl_elem_add_info { +struct snd_ctl_elem_id id;/* element ID */ +snd_ctl_elem_type_t type;/* value type */ +unsigned int access;/* value access (bitmask) */ +unsigned int count;/* count of values */ +__kernel_pid_t elem_count;/* # of elements to create (0=1) */ +union { +struct { +long min;/* R: minimum value */ +long max;/* R: maximum value */ +long step;/* R: step (0 variable) */ + } integer; +struct { +long long min;/* R: minimum value */ +long long max;/* R: maximum value */ +long long step;/* R: step (0 variable) */ + } integer64; +unsigned char reserved[128 ]; + } value; +unsigned char reserved[64 ]; +}; +struct snd_ctl_elem_value {struct snd_ctl_elem_id id;/* W: element ID */ unsigned int indirect:1 ;/* W: indirect access - obsoleted */ @@ -854,8 +881,8 @@ struct snd_ctl_tlv {#define SNDRV_CTL_IOCTL_ELEM_LOCK _IOW('U', 0x14, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_ELEM_UNLOCK _IOW('U', 0x15, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_SUBSCRIBE_EVENTS _IOWR('U', 0x16, int) -#define SNDRV_CTL_IOCTL_ELEM_ADD _IOWR('U', 0x17, struct snd_ctl_elem_info) -#define SNDRV_CTL_IOCTL_ELEM_REPLACE _IOWR('U', 0x18, struct snd_ctl_elem_info) +#define SNDRV_CTL_IOCTL_ELEM_ADD _IOWR('U', 0x17, struct snd_ctl_elem_add_info) +#define SNDRV_CTL_IOCTL_ELEM_REPLACE _IOWR('U', 0x18, struct snd_ctl_elem_add_info)#define SNDRV_CTL_IOCTL_ELEM_REMOVE _IOWR('U', 0x19, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_TLV_READ _IOWR('U', 0x1a, struct snd_ctl_tlv)#define SNDRV_CTL_IOCTL_TLV_WRITE _IOWR('U', 0x1b, struct snd_ctl_tlv)sound/core/control.c +1
-1@@ -1064,7 +1064,7 @@ static void snd_ctl_elem_user_free(struct snd_kcontrol *kcontrol) }static int snd_ctl_elem_add (struct snd_ctl_file *file, -struct snd_ctl_elem_info *info,int replace) +struct snd_ctl_elem_add_info *info,int replace) {struct snd_card *card = file->card;struct snd_kcontrol kctl, *_kctl;
sound/core/control.c static int snd_ctl_elem_add(struct snd_ctl_file *file, · line 1102
Quoted 5 lines
sound/core/control.c +1
-1@@ -1099,7 +1099,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,if (err <0 )return err;memcpy (&kctl.id, &info->id,sizeof (info->id)); - kctl.count = info->owner ? info->owner :1 ; + kctl.count = info->elem_count ? info->elem_count :1 ;
Hi Takashi,
I don't how this approach can solve this problem. There's still an implicit assumption here that user space application will clear info->elem_count to zero before it issues the REPLACE ioctl, right?
sound/core/control.c static int snd_ctl_elem_add(struct snd_ctl_file *file, · line 1142
Quoted 9 lines
sound/core/control.c +1
-1access |=SNDRV_CTL_ELEM_ACCESS_USER ; kctl.info = snd_ctl_elem_user_info;if (access &SNDRV_CTL_ELEM_ACCESS_READ )@@ -1139,7 +1139,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file, ue =kzalloc (sizeof (struct user_element ) + private_size,GFP_KERNEL );if (ue ==NULL )return -ENOMEM ; - ue->info = *info; +memcpy (&ue->info, info,sizeof (*info));
Should we add some assertion here checking the size of these two structure should be the same?
Quoted 16 lines
sound/core/control.c +3
-2ue->info.access =0 ; ue->elem_data = (char *)ue +sizeof (*ue); ue->elem_data_size = private_size;@@ -1164,9 +1164,10 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file, }static int snd_ctl_elem_add_user (struct snd_ctl_file *file, -struct snd_ctl_elem_info __user *_info,int replace) +struct snd_ctl_elem_add_info __user *_info, +int replace) { -struct snd_ctl_elem_info info; +struct snd_ctl_elem_add_info info;if (copy_from_user (&info, _info,sizeof (info)))return -EFAULT ;return snd_ctl_elem_add (file, &info, replace);
Signature · 1 line
guanqun
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
At Wed, 24 Aug 2011 19:46:13 +0800,
Quoted 74 lines · Lu Guanqun
Quoted 64 lines · Takashi Iwai
---
include/sound/asound.h +29
-2@@ -813,6 +813,33 @@ struct snd_ctl_elem_info {unsigned char reserved[64 -4 *sizeof (unsigned short )]; }; +/* The struct used for ELEM_ADD and ELEM_REPLACE ioctls; +* This is is identical with snd_ctl_elem_info, but just owner field is +* different, keeping the number of elements to be created. +* The unused fields are stripped for simplicity. +*/ +struct snd_ctl_elem_add_info { +struct snd_ctl_elem_id id;/* element ID */ +snd_ctl_elem_type_t type;/* value type */ +unsigned int access;/* value access (bitmask) */ +unsigned int count;/* count of values */ +__kernel_pid_t elem_count;/* # of elements to create (0=1) */ +union { +struct { +long min;/* R: minimum value */ +long max;/* R: maximum value */ +long step;/* R: step (0 variable) */ + } integer; +struct { +long long min;/* R: minimum value */ +long long max;/* R: maximum value */ +long long step;/* R: step (0 variable) */ + } integer64; +unsigned char reserved[128 ]; + } value; +unsigned char reserved[64 ]; +}; +struct snd_ctl_elem_value {struct snd_ctl_elem_id id;/* W: element ID */ unsigned int indirect:1 ;/* W: indirect access - obsoleted */ @@ -854,8 +881,8 @@ struct snd_ctl_tlv {#define SNDRV_CTL_IOCTL_ELEM_LOCK _IOW('U', 0x14, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_ELEM_UNLOCK _IOW('U', 0x15, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_SUBSCRIBE_EVENTS _IOWR('U', 0x16, int) -#define SNDRV_CTL_IOCTL_ELEM_ADD _IOWR('U', 0x17, struct snd_ctl_elem_info) -#define SNDRV_CTL_IOCTL_ELEM_REPLACE _IOWR('U', 0x18, struct snd_ctl_elem_info) +#define SNDRV_CTL_IOCTL_ELEM_ADD _IOWR('U', 0x17, struct snd_ctl_elem_add_info) +#define SNDRV_CTL_IOCTL_ELEM_REPLACE _IOWR('U', 0x18, struct snd_ctl_elem_add_info)#define SNDRV_CTL_IOCTL_ELEM_REMOVE _IOWR('U', 0x19, struct snd_ctl_elem_id)#define SNDRV_CTL_IOCTL_TLV_READ _IOWR('U', 0x1a, struct snd_ctl_tlv)#define SNDRV_CTL_IOCTL_TLV_WRITE _IOWR('U', 0x1b, struct snd_ctl_tlv)sound/core/control.c +1
-1@@ -1064,7 +1064,7 @@ static void snd_ctl_elem_user_free(struct snd_kcontrol *kcontrol) }static int snd_ctl_elem_add (struct snd_ctl_file *file, -struct snd_ctl_elem_info *info,int replace) +struct snd_ctl_elem_add_info *info,int replace) {struct snd_card *card = file->card;struct snd_kcontrol kctl, *_kctl;sound/core/control.c static int snd_ctl_elem_add(struct snd_ctl_file *file, · line 1102
Quoted 5 lines
sound/core/control.c +1
-1@@ -1099,7 +1099,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,if (err <0 )return err;memcpy (&kctl.id, &info->id,sizeof (info->id)); - kctl.count = info->owner ? info->owner :1 ; + kctl.count = info->elem_count ? info->elem_count :1 ;Hi Takashi,
Quoted 1 line
I don't how this approach can solve this problem.
There is no problem. You can't copy the data from ELEM_INFO for ELEM_ADD/REPLACE. It was a wrong assumption to reuse the data. Instead, you had to set up the struct field manually. That is, the test procedure was simply wrong.
Quoted 3 lines
There's still an
implicit assumption here that user space application will clearinfo->elem_countto zero before it issues the REPLACE ioctl, right?
Yes, the point is that it's no longer same structure. So, you can't copy from the result from ELEM_INFO ioctl any more. Instead, you need to set up the struct field manually.
Quoted 4 lines
Quoted 4 lines
sound/core/control.c +0
-0access |=SNDRV_CTL_ELEM_ACCESS_USER ; kctl.info = snd_ctl_elem_user_info;if (access &SNDRV_CTL_ELEM_ACCESS_READ )@@ -1139,7 +1139,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,
Quoted 8 lines
sound/core/control.c static int snd_ctl_elem_add(struct snd_ctl_file *file, · line 1142
Quoted 5 lines
sound/core/control.c +1
-1ue =kzalloc (sizeof (struct user_element ) + private_size,GFP_KERNEL );if (ue ==NULL )return -ENOMEM ; - ue->info = *info; +memcpy (&ue->info, info,sizeof (*info));Should we add some assertion here checking the size of these two structure should be the same?
It could, at most by BUILD_BUG_ON().
thanks,
Takashi
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
Quoted 9 lines · Takashi Iwai
Quoted 3 lines
Hi Takashi,
I don't how this approach can solve this problem.
There is no problem. You can't copy the data from
ELEM_INFOfor ELEM_ADD/REPLACE. It was a wrong assumption to reuse the data. Instead, you had to set up the struct field manually. That is, the test procedure was simply wrong.
Quoted 7 lines
Quoted 3 lines
There's still an
implicit assumption here that user space application will clearinfo->elem_countto zero before it issues the REPLACE ioctl, right?Yes, the point is that it's no longer same structure. So, you can't copy from the result from
ELEM_INFOioctl any more. Instead, you need to set up the struct field manually.
Thanks Takashi, that's the info I'm looking for. We need to state this requirements explicitly, it seems we lack these kind of documentation and thus it results in poorly written test cases in user space.
Is there any man pages or documentation we can state this clearly? I can help to contribute some documentation or even add some sample programs there.
Signature · 1 line
guanqun
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
At Wed, 24 Aug 2011 20:46:38 +0800,
Quoted 15 lines · Lu Guanqun
Quoted 13 lines · Takashi Iwai
Quoted 3 lines
Hi Takashi,
I don't how this approach can solve this problem.
There is no problem. You can't copy the data from
ELEM_INFOfor ELEM_ADD/REPLACE. It was a wrong assumption to reuse the data. Instead, you had to set up the struct field manually. That is, the test procedure was simply wrong.Quoted 3 lines
There's still an
implicit assumption here that user space application will clearinfo->elem_countto zero before it issues the REPLACE ioctl, right?
Quoted 7 lines
Quoted 3 lines
Yes, the point is that it's no longer same structure. So, you can't copy from the result from
ELEM_INFOioctl any more. Instead, you need to set up the struct field manually.Thanks Takashi, that's the info I'm looking for. We need to state this requirements explicitly, it seems we lack these kind of documentation and thus it results in poorly written test cases in user space.
Yep.
Quoted 1 line
Is there any man pages or documentation we can state this clearly?
No. Partly because the user-space apps are supposed to use alsa-lib.
Quoted 3 lines
I can
help to contribute some documentation or even add some sample programs there.
Yes, please, that'll be helpful.
thanks,
Takashi
Re: [PATCH 3/3] ALSA: core: set kcontrol's count field correctly
In reply to Takashi IwaiAt Wed, 24 Aug 2011 08:06:34 +0200,
Quoted 13 lines · Takashi Iwai
At Wed, 24 Aug 2011 11:12:43 +0800,
Quoted 10 lines · Lu Guanqun
I don't see how info's owner field relates to kcontrol's count field. It should assign to info's count instead.
Let's assume this scenario:
1. user reads the control element from kernel (the owner field is set)
2. user changes some values
3. user issues 'SNDRV_CTL_IOCTL_ELEM_REPLACE' ioctl.
Quoted 7 lines
Quoted 2 lines
With the original code, 'kctl' here gets a large count number due to its non-empty owner field. Therefore it fails on subsequent call
snd_ctl_new().In
IOCTL_ELEM_ADDand REPLACE, count and owner struct fields have different meanings from others. The count contains the array size, not the number of identical elements. And the owner contains the number of elements to be created.
Correction: the count field means same as others. The info.count specifies the array size in IOCTL_ELEM_INFO, too.
The only difference is an (ab)use of owner field.
Takashi
At Wed, 24 Aug 2011 11:12:34 +0800,
Quoted 4 lines · Lu Guanqun
remove trailing tab on the line.
Signed-off-by Lu Guanqun <[email protected]>
Applied now. Thanks.
Takashi
Quoted 18 lines
--- sound/core/control.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)sound/core/control.c +1
-1@@ -1072,7 +1072,7 @@ static int snd_ctl_elem_add(struct snd_ctl_file *file,long private_size;struct user_element *ue;int idx, err; - +if (card->user_ctl_count >=MAX_USER_CONTROLS )return -ENOMEM ;if (info->count <1 )