* [PATCH] agp: remove read/write stubs
@ 2014-06-15 21:02 Mathias Krause
2014-07-02 6:15 ` Mathias Krause
0 siblings, 1 reply; 7+ messages in thread
From: Mathias Krause @ 2014-06-15 21:02 UTC (permalink / raw)
To: David Airlie; +Cc: linux-kernel, Mathias Krause, Alexander Viro
The VFS layer handles those in the very same way, if unset. No need for
additional stubs.
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
---
drivers/char/agp/frontend.c | 15 ---------------
1 file changed, 15 deletions(-)
diff --git a/drivers/char/agp/frontend.c b/drivers/char/agp/frontend.c
index b29703324e..09f17eb734 100644
--- a/drivers/char/agp/frontend.c
+++ b/drivers/char/agp/frontend.c
@@ -710,19 +710,6 @@ static int agp_open(struct inode *inode, struct file *file)
return 0;
}
-
-static ssize_t agp_read(struct file *file, char __user *buf,
- size_t count, loff_t * ppos)
-{
- return -EINVAL;
-}
-
-static ssize_t agp_write(struct file *file, const char __user *buf,
- size_t count, loff_t * ppos)
-{
- return -EINVAL;
-}
-
static int agpioc_info_wrap(struct agp_file_private *priv, void __user *arg)
{
struct agp_info userinfo;
@@ -1047,8 +1034,6 @@ static const struct file_operations agp_fops =
{
.owner = THIS_MODULE,
.llseek = no_llseek,
- .read = agp_read,
- .write = agp_write,
.unlocked_ioctl = agp_ioctl,
#ifdef CONFIG_COMPAT
.compat_ioctl = compat_agp_ioctl,
--
1.7.10.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH] agp: remove read/write stubs
2014-06-15 21:02 [PATCH] agp: remove read/write stubs Mathias Krause
@ 2014-07-02 6:15 ` Mathias Krause
2014-07-08 22:48 ` Dave Airlie
0 siblings, 1 reply; 7+ messages in thread
From: Mathias Krause @ 2014-07-02 6:15 UTC (permalink / raw)
To: David Airlie; +Cc: linux-kernel, Mathias Krause, Alexander Viro
On 15 June 2014 23:02, Mathias Krause <minipli@googlemail.com> wrote:
> The VFS layer handles those in the very same way, if unset. No need for
> additional stubs.
>
> Signed-off-by: Mathias Krause <minipli@googlemail.com>
> Cc: Alexander Viro <viro@zeniv.linux.org.uk>
> ---
> drivers/char/agp/frontend.c | 15 ---------------
> 1 file changed, 15 deletions(-)
>
> diff --git a/drivers/char/agp/frontend.c b/drivers/char/agp/frontend.c
> index b29703324e..09f17eb734 100644
> --- a/drivers/char/agp/frontend.c
> +++ b/drivers/char/agp/frontend.c
> @@ -710,19 +710,6 @@ static int agp_open(struct inode *inode, struct file *file)
> return 0;
> }
>
> -
> -static ssize_t agp_read(struct file *file, char __user *buf,
> - size_t count, loff_t * ppos)
> -{
> - return -EINVAL;
> -}
> -
> -static ssize_t agp_write(struct file *file, const char __user *buf,
> - size_t count, loff_t * ppos)
> -{
> - return -EINVAL;
> -}
> -
> static int agpioc_info_wrap(struct agp_file_private *priv, void __user *arg)
> {
> struct agp_info userinfo;
> @@ -1047,8 +1034,6 @@ static const struct file_operations agp_fops =
> {
> .owner = THIS_MODULE,
> .llseek = no_llseek,
> - .read = agp_read,
> - .write = agp_write,
> .unlocked_ioctl = agp_ioctl,
> #ifdef CONFIG_COMPAT
> .compat_ioctl = compat_agp_ioctl,
> --
> 1.7.10.4
>
Ping. Any objections, comments? David?
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] agp: remove read/write stubs
2014-07-02 6:15 ` Mathias Krause
@ 2014-07-08 22:48 ` Dave Airlie
2014-07-09 6:52 ` Mathias Krause
0 siblings, 1 reply; 7+ messages in thread
From: Dave Airlie @ 2014-07-08 22:48 UTC (permalink / raw)
To: Mathias Krause; +Cc: David Airlie, linux-kernel, Alexander Viro
On 2 July 2014 16:15, Mathias Krause <minipli@googlemail.com> wrote:
> On 15 June 2014 23:02, Mathias Krause <minipli@googlemail.com> wrote:
>> The VFS layer handles those in the very same way, if unset. No need for
>> additional stubs.
>>
>> Signed-off-by: Mathias Krause <minipli@googlemail.com>
>> Cc: Alexander Viro <viro@zeniv.linux.org.uk>
>> ---
>> drivers/char/agp/frontend.c | 15 ---------------
>> 1 file changed, 15 deletions(-)
>>
>> diff --git a/drivers/char/agp/frontend.c b/drivers/char/agp/frontend.c
>> index b29703324e..09f17eb734 100644
>> --- a/drivers/char/agp/frontend.c
>> +++ b/drivers/char/agp/frontend.c
>> @@ -710,19 +710,6 @@ static int agp_open(struct inode *inode, struct file *file)
>> return 0;
>> }
>>
>> -
>> -static ssize_t agp_read(struct file *file, char __user *buf,
>> - size_t count, loff_t * ppos)
>> -{
>> - return -EINVAL;
>> -}
>> -
>> -static ssize_t agp_write(struct file *file, const char __user *buf,
>> - size_t count, loff_t * ppos)
>> -{
>> - return -EINVAL;
>> -}
>> -
>> static int agpioc_info_wrap(struct agp_file_private *priv, void __user *arg)
>> {
>> struct agp_info userinfo;
>> @@ -1047,8 +1034,6 @@ static const struct file_operations agp_fops =
>> {
>> .owner = THIS_MODULE,
>> .llseek = no_llseek,
>> - .read = agp_read,
>> - .write = agp_write,
>> .unlocked_ioctl = agp_ioctl,
>> #ifdef CONFIG_COMPAT
>> .compat_ioctl = compat_agp_ioctl,
>> --
>> 1.7.10.4
>>
>
> Ping. Any objections, comments? David?
Are you sure VFS handles this the same way? can you point me where it
checks ->read and returns -EINVAL, I'm probably just missing it in my
quick look over fs/read_write.c
Dave.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] agp: remove read/write stubs
2014-07-08 22:48 ` Dave Airlie
@ 2014-07-09 6:52 ` Mathias Krause
2014-07-22 19:50 ` Mathias Krause
0 siblings, 1 reply; 7+ messages in thread
From: Mathias Krause @ 2014-07-09 6:52 UTC (permalink / raw)
To: Dave Airlie; +Cc: David Airlie, linux-kernel, Alexander Viro
On 9 July 2014 00:48, Dave Airlie <airlied@gmail.com> wrote:
> On 2 July 2014 16:15, Mathias Krause <minipli@googlemail.com> wrote:
>> On 15 June 2014 23:02, Mathias Krause <minipli@googlemail.com> wrote:
>>> The VFS layer handles those in the very same way, if unset. No need for
>>> additional stubs.
>>>
>>> Signed-off-by: Mathias Krause <minipli@googlemail.com>
>>> Cc: Alexander Viro <viro@zeniv.linux.org.uk>
>>> ---
>>> drivers/char/agp/frontend.c | 15 ---------------
>>> 1 file changed, 15 deletions(-)
>>>
>>> diff --git a/drivers/char/agp/frontend.c b/drivers/char/agp/frontend.c
>>> index b29703324e..09f17eb734 100644
>>> --- a/drivers/char/agp/frontend.c
>>> +++ b/drivers/char/agp/frontend.c
>>> @@ -710,19 +710,6 @@ static int agp_open(struct inode *inode, struct file *file)
>>> return 0;
>>> }
>>>
>>> -
>>> -static ssize_t agp_read(struct file *file, char __user *buf,
>>> - size_t count, loff_t * ppos)
>>> -{
>>> - return -EINVAL;
>>> -}
>>> -
>>> -static ssize_t agp_write(struct file *file, const char __user *buf,
>>> - size_t count, loff_t * ppos)
>>> -{
>>> - return -EINVAL;
>>> -}
>>> -
>>> static int agpioc_info_wrap(struct agp_file_private *priv, void __user *arg)
>>> {
>>> struct agp_info userinfo;
>>> @@ -1047,8 +1034,6 @@ static const struct file_operations agp_fops =
>>> {
>>> .owner = THIS_MODULE,
>>> .llseek = no_llseek,
>>> - .read = agp_read,
>>> - .write = agp_write,
>>> .unlocked_ioctl = agp_ioctl,
>>> #ifdef CONFIG_COMPAT
>>> .compat_ioctl = compat_agp_ioctl,
>>> --
>>> 1.7.10.4
>>>
>>
>> Ping. Any objections, comments? David?
>
> Are you sure VFS handles this the same way? can you point me where it
> checks ->read and returns -EINVAL, I'm probably just missing it in my
> quick look over fs/read_write.c
The VFS code changed during the last merge window so the ->read and
->write tests are no longer directly visible. The presence of a ->read
/ ->write function is now remembered at open(2) time and stored in
->f_mode as a flag.
The corresponding commit for the VFS change is 7f7f25e82d "replace
checking for ->read/->aio_read presence with check in ->f_mode".
Despite its subject, it also handles ->write / ->aio_write. The flags
are set in fs/open:do_dentry_open():
if ((f->f_mode & FMODE_READ) &&
likely(f->f_op->read || f->f_op->aio_read || f->f_op->read_iter))
f->f_mode |= FMODE_CAN_READ;
if ((f->f_mode & FMODE_WRITE) &&
likely(f->f_op->write || f->f_op->aio_write || f->f_op->write_iter))
f->f_mode |= FMODE_CAN_WRITE;
With that change in mind, the redundancy of the agp_read() and
agp_write() functions can be seen for read(2) in vfs_read():
if (!(file->f_mode & FMODE_CAN_READ))
return -EINVAL;
..and for write(2) in vfs_write():
if (!(file->f_mode & FMODE_CAN_WRITE))
return -EINVAL;
The other entry points (readv / writev / compat wrappers) have been
changed accordingly, so those handle a missing ->read / ->write
callback in the same way: returning -EINVAL.
Regards,
Mathias
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] agp: remove read/write stubs
2014-07-09 6:52 ` Mathias Krause
@ 2014-07-22 19:50 ` Mathias Krause
2014-08-01 20:01 ` Mathias Krause
0 siblings, 1 reply; 7+ messages in thread
From: Mathias Krause @ 2014-07-22 19:50 UTC (permalink / raw)
To: Dave Airlie; +Cc: David Airlie, linux-kernel, Alexander Viro
On 9 July 2014 08:52, Mathias Krause <minipli@googlemail.com> wrote:
> On 9 July 2014 00:48, Dave Airlie <airlied@gmail.com> wrote:
>> On 2 July 2014 16:15, Mathias Krause <minipli@googlemail.com> wrote:
>>> On 15 June 2014 23:02, Mathias Krause <minipli@googlemail.com> wrote:
>>>> The VFS layer handles those in the very same way, if unset. No need for
>>>> additional stubs.
>>>>
>>>> Signed-off-by: Mathias Krause <minipli@googlemail.com>
>>>> Cc: Alexander Viro <viro@zeniv.linux.org.uk>
>>>> ---
>>>> drivers/char/agp/frontend.c | 15 ---------------
>>>> 1 file changed, 15 deletions(-)
>>>>
>>>> diff --git a/drivers/char/agp/frontend.c b/drivers/char/agp/frontend.c
>>>> index b29703324e..09f17eb734 100644
>>>> --- a/drivers/char/agp/frontend.c
>>>> +++ b/drivers/char/agp/frontend.c
>>>> @@ -710,19 +710,6 @@ static int agp_open(struct inode *inode, struct file *file)
>>>> return 0;
>>>> }
>>>>
>>>> -
>>>> -static ssize_t agp_read(struct file *file, char __user *buf,
>>>> - size_t count, loff_t * ppos)
>>>> -{
>>>> - return -EINVAL;
>>>> -}
>>>> -
>>>> -static ssize_t agp_write(struct file *file, const char __user *buf,
>>>> - size_t count, loff_t * ppos)
>>>> -{
>>>> - return -EINVAL;
>>>> -}
>>>> -
>>>> static int agpioc_info_wrap(struct agp_file_private *priv, void __user *arg)
>>>> {
>>>> struct agp_info userinfo;
>>>> @@ -1047,8 +1034,6 @@ static const struct file_operations agp_fops =
>>>> {
>>>> .owner = THIS_MODULE,
>>>> .llseek = no_llseek,
>>>> - .read = agp_read,
>>>> - .write = agp_write,
>>>> .unlocked_ioctl = agp_ioctl,
>>>> #ifdef CONFIG_COMPAT
>>>> .compat_ioctl = compat_agp_ioctl,
>>>> --
>>>> 1.7.10.4
>>>>
>>>
>>> Ping. Any objections, comments? David?
>>
>> Are you sure VFS handles this the same way? can you point me where it
>> checks ->read and returns -EINVAL, I'm probably just missing it in my
>> quick look over fs/read_write.c
>
> The VFS code changed during the last merge window so the ->read and
> ->write tests are no longer directly visible. The presence of a ->read
> / ->write function is now remembered at open(2) time and stored in
> ->f_mode as a flag.
>
> The corresponding commit for the VFS change is 7f7f25e82d "replace
> checking for ->read/->aio_read presence with check in ->f_mode".
> Despite its subject, it also handles ->write / ->aio_write. The flags
> are set in fs/open:do_dentry_open():
>
> if ((f->f_mode & FMODE_READ) &&
> likely(f->f_op->read || f->f_op->aio_read || f->f_op->read_iter))
> f->f_mode |= FMODE_CAN_READ;
> if ((f->f_mode & FMODE_WRITE) &&
> likely(f->f_op->write || f->f_op->aio_write || f->f_op->write_iter))
> f->f_mode |= FMODE_CAN_WRITE;
>
> With that change in mind, the redundancy of the agp_read() and
> agp_write() functions can be seen for read(2) in vfs_read():
>
> if (!(file->f_mode & FMODE_CAN_READ))
> return -EINVAL;
>
> ..and for write(2) in vfs_write():
>
> if (!(file->f_mode & FMODE_CAN_WRITE))
> return -EINVAL;
>
> The other entry points (readv / writev / compat wrappers) have been
> changed accordingly, so those handle a missing ->read / ->write
> callback in the same way: returning -EINVAL.
>
> Regards,
> Mathias
Ping. Any more comments, David?
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] agp: remove read/write stubs
2014-07-22 19:50 ` Mathias Krause
@ 2014-08-01 20:01 ` Mathias Krause
2014-08-02 10:08 ` Dave Airlie
0 siblings, 1 reply; 7+ messages in thread
From: Mathias Krause @ 2014-08-01 20:01 UTC (permalink / raw)
To: Dave Airlie; +Cc: David Airlie, linux-kernel, Alexander Viro
On 22 July 2014 21:50, Mathias Krause <minipli@googlemail.com> wrote:
> On 9 July 2014 08:52, Mathias Krause <minipli@googlemail.com> wrote:
>> On 9 July 2014 00:48, Dave Airlie <airlied@gmail.com> wrote:
>>> On 2 July 2014 16:15, Mathias Krause <minipli@googlemail.com> wrote:
>>>> On 15 June 2014 23:02, Mathias Krause <minipli@googlemail.com> wrote:
>>>>> The VFS layer handles those in the very same way, if unset. No need for
>>>>> additional stubs.
>>>>>
>>>>> Signed-off-by: Mathias Krause <minipli@googlemail.com>
>>>>> Cc: Alexander Viro <viro@zeniv.linux.org.uk>
>>>>> ---
>>>>> drivers/char/agp/frontend.c | 15 ---------------
>>>>> 1 file changed, 15 deletions(-)
>>>>>
>>>>> diff --git a/drivers/char/agp/frontend.c b/drivers/char/agp/frontend.c
>>>>> index b29703324e..09f17eb734 100644
>>>>> --- a/drivers/char/agp/frontend.c
>>>>> +++ b/drivers/char/agp/frontend.c
>>>>> @@ -710,19 +710,6 @@ static int agp_open(struct inode *inode, struct file *file)
>>>>> return 0;
>>>>> }
>>>>>
>>>>> -
>>>>> -static ssize_t agp_read(struct file *file, char __user *buf,
>>>>> - size_t count, loff_t * ppos)
>>>>> -{
>>>>> - return -EINVAL;
>>>>> -}
>>>>> -
>>>>> -static ssize_t agp_write(struct file *file, const char __user *buf,
>>>>> - size_t count, loff_t * ppos)
>>>>> -{
>>>>> - return -EINVAL;
>>>>> -}
>>>>> -
>>>>> static int agpioc_info_wrap(struct agp_file_private *priv, void __user *arg)
>>>>> {
>>>>> struct agp_info userinfo;
>>>>> @@ -1047,8 +1034,6 @@ static const struct file_operations agp_fops =
>>>>> {
>>>>> .owner = THIS_MODULE,
>>>>> .llseek = no_llseek,
>>>>> - .read = agp_read,
>>>>> - .write = agp_write,
>>>>> .unlocked_ioctl = agp_ioctl,
>>>>> #ifdef CONFIG_COMPAT
>>>>> .compat_ioctl = compat_agp_ioctl,
>>>>> --
>>>>> 1.7.10.4
>>>>>
>>>>
>>>> Ping. Any objections, comments? David?
>>>
>>> Are you sure VFS handles this the same way? can you point me where it
>>> checks ->read and returns -EINVAL, I'm probably just missing it in my
>>> quick look over fs/read_write.c
>>
>> The VFS code changed during the last merge window so the ->read and
>> ->write tests are no longer directly visible. The presence of a ->read
>> / ->write function is now remembered at open(2) time and stored in
>> ->f_mode as a flag.
>>
>> The corresponding commit for the VFS change is 7f7f25e82d "replace
>> checking for ->read/->aio_read presence with check in ->f_mode".
>> Despite its subject, it also handles ->write / ->aio_write. The flags
>> are set in fs/open:do_dentry_open():
>>
>> if ((f->f_mode & FMODE_READ) &&
>> likely(f->f_op->read || f->f_op->aio_read || f->f_op->read_iter))
>> f->f_mode |= FMODE_CAN_READ;
>> if ((f->f_mode & FMODE_WRITE) &&
>> likely(f->f_op->write || f->f_op->aio_write || f->f_op->write_iter))
>> f->f_mode |= FMODE_CAN_WRITE;
>>
>> With that change in mind, the redundancy of the agp_read() and
>> agp_write() functions can be seen for read(2) in vfs_read():
>>
>> if (!(file->f_mode & FMODE_CAN_READ))
>> return -EINVAL;
>>
>> ..and for write(2) in vfs_write():
>>
>> if (!(file->f_mode & FMODE_CAN_WRITE))
>> return -EINVAL;
>>
>> The other entry points (readv / writev / compat wrappers) have been
>> changed accordingly, so those handle a missing ->read / ->write
>> callback in the same way: returning -EINVAL.
>>
>> Regards,
>> Mathias
>
> Ping. Any more comments, David?
Ping, ping. David, did you had time to look at the patch again?
Regards,
Mathias
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] agp: remove read/write stubs
2014-08-01 20:01 ` Mathias Krause
@ 2014-08-02 10:08 ` Dave Airlie
0 siblings, 0 replies; 7+ messages in thread
From: Dave Airlie @ 2014-08-02 10:08 UTC (permalink / raw)
To: Mathias Krause; +Cc: Dave Airlie, linux-kernel, Alexander Viro
> >
> > Ping. Any more comments, David?
>
> Ping, ping. David, did you had time to look at the patch again?
Merged finally thanks for reminding me, agp is kinda a dead zone.
Dave.
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2014-08-02 10:08 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-06-15 21:02 [PATCH] agp: remove read/write stubs Mathias Krause
2014-07-02 6:15 ` Mathias Krause
2014-07-08 22:48 ` Dave Airlie
2014-07-09 6:52 ` Mathias Krause
2014-07-22 19:50 ` Mathias Krause
2014-08-01 20:01 ` Mathias Krause
2014-08-02 10:08 ` Dave Airlie
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).