* [GIT] /proc/uptime fix
@ 2008-11-03 14:50 Alexey Dobriyan
2008-11-03 18:14 ` Linus Torvalds
0 siblings, 1 reply; 3+ messages in thread
From: Alexey Dobriyan @ 2008-11-03 14:50 UTC (permalink / raw)
To: torvalds; +Cc: linux-kernel
Linus, please, pull from
git://git.kernel.org/pub/scm/linux/kernel/git/adobriyan/proc.git proc-linus
to get pread("/proc/uptime") back for some VMware userspace:
fs/proc/uptime.c | 38 ++++++++++++++++++++------------------
1 files changed, 20 insertions(+), 18 deletions(-)
Alexey Dobriyan (1):
proc: revert /proc/uptime to ->read_proc hook
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [GIT] /proc/uptime fix
2008-11-03 14:50 [GIT] /proc/uptime fix Alexey Dobriyan
@ 2008-11-03 18:14 ` Linus Torvalds
2008-11-03 20:25 ` Alexey Dobriyan
0 siblings, 1 reply; 3+ messages in thread
From: Linus Torvalds @ 2008-11-03 18:14 UTC (permalink / raw)
To: Alexey Dobriyan, Al Viro; +Cc: Linux Kernel Mailing List
On Mon, 3 Nov 2008, Alexey Dobriyan wrote:
>
> Alexey Dobriyan (1):
> proc: revert /proc/uptime to ->read_proc hook
Ok, pulled, but it's a bit sad.
Al: the commit is 6c87df37dcb9c6c33923707fa5191e0a65874d60, and the
message explains it all:
Turned out some VMware userspace does pread(2) on /proc/uptime, but
seqfiles currently don't allow pread() resulting in -ESPIPE.
Seqfiles in theory can do pread(), but this can be a long story,
so revert to ->read_proc until then.
and I bet that the pread() usage is just with a constant loff_t of 0, and
just there to avoid a simple case of "lseek+read".
I wonder how hard it is to make seqfiles support pread. Maybe it's
something as trivial as just flushing the buffer whenever 'pos' doesn't
match f_pos?
ie something like the following (TOTALLY UNTESTED!)
I may be way off base here, Al needs to whack some sense into this.
Linus
---
fs/seq_file.c | 8 ++++++--
1 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/fs/seq_file.c b/fs/seq_file.c
index eba2eab..eb95ec9 100644
--- a/fs/seq_file.c
+++ b/fs/seq_file.c
@@ -48,8 +48,8 @@ int seq_open(struct file *file, const struct seq_operations *op)
*/
file->f_version = 0;
- /* SEQ files support lseek, but not pread/pwrite */
- file->f_mode &= ~(FMODE_PREAD | FMODE_PWRITE);
+ /* SEQ files support lseek, but not pwrite */
+ file->f_mode &= ~(FMODE_PWRITE);
return 0;
}
EXPORT_SYMBOL(seq_open);
@@ -91,6 +91,8 @@ ssize_t seq_read(struct file *file, char __user *buf, size_t size, loff_t *ppos)
if (!m->buf)
goto Enomem;
}
+ if (pos != file->f_pos)
+ m->count = 0;
/* if not empty - flush it first */
if (m->count) {
n = min(m->count, size);
@@ -175,6 +177,8 @@ Done:
else
*ppos += copied;
file->f_version = m->version;
+ if (pos != file->f_pos)
+ m->count = 0;
mutex_unlock(&m->lock);
return copied;
Enomem:
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [GIT] /proc/uptime fix
2008-11-03 18:14 ` Linus Torvalds
@ 2008-11-03 20:25 ` Alexey Dobriyan
0 siblings, 0 replies; 3+ messages in thread
From: Alexey Dobriyan @ 2008-11-03 20:25 UTC (permalink / raw)
To: Linus Torvalds; +Cc: Al Viro, linux-kernel
On Mon, Nov 03, 2008 at 10:14:16AM -0800, Linus Torvalds wrote:
> On Mon, 3 Nov 2008, Alexey Dobriyan wrote:
> >
> > Alexey Dobriyan (1):
> > proc: revert /proc/uptime to ->read_proc hook
>
> Ok, pulled, but it's a bit sad.
>
> Al: the commit is 6c87df37dcb9c6c33923707fa5191e0a65874d60, and the
> message explains it all:
>
> Turned out some VMware userspace does pread(2) on /proc/uptime, but
> seqfiles currently don't allow pread() resulting in -ESPIPE.
>
> Seqfiles in theory can do pread(), but this can be a long story,
> so revert to ->read_proc until then.
>
> and I bet that the pread() usage is just with a constant loff_t of 0, and
> just there to avoid a simple case of "lseek+read".
>
> I wonder how hard it is to make seqfiles support pread.
Not hard at all, if ->read hook has information if it's called from read(2)
or pread(2) (or something equivalent).
> Maybe it's something as trivial as just flushing the buffer whenever 'pos'
> doesn't match f_pos?
> --- a/fs/seq_file.c
> +++ b/fs/seq_file.c
> @@ -48,8 +48,8 @@ int seq_open(struct file *file, const struct seq_operations *op)
> */
> file->f_version = 0;
>
> - /* SEQ files support lseek, but not pread/pwrite */
> - file->f_mode &= ~(FMODE_PREAD | FMODE_PWRITE);
> + /* SEQ files support lseek, but not pwrite */
> + file->f_mode &= ~(FMODE_PWRITE);
FMODE_PWRITE is aliased to FMODE_PREAD so you have to split them first. ;-)
> @@ -91,6 +91,8 @@ ssize_t seq_read(struct file *file, char __user *buf, size_t size, loff_t *ppos)
> if (!m->buf)
> goto Enomem;
> }
> + if (pos != file->f_pos)
> + m->count = 0;
> /* if not empty - flush it first */
> if (m->count) {
> n = min(m->count, size);
> @@ -175,6 +177,8 @@ Done:
> else
> *ppos += copied;
> file->f_version = m->version;
> + if (pos != file->f_pos)
> + m->count = 0;
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2008-11-03 20:32 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-11-03 14:50 [GIT] /proc/uptime fix Alexey Dobriyan
2008-11-03 18:14 ` Linus Torvalds
2008-11-03 20:25 ` Alexey Dobriyan
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®