2006-09-08 16:39:38

by Cédric Le Goater

[permalink] [raw]
Subject: [patch -mm] update mq_notify to use a struct pid

message queues can signal a process waiting for a message.

this patch replaces the pid_t value with a struct pid to avoid pid wrap
around problems.

Signed-off-by: Cedric Le Goater <[email protected]>
Cc: Eric Biederman <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: [email protected]

---
ipc/mqueue.c | 27 +++++++++++++++------------
1 file changed, 15 insertions(+), 12 deletions(-)

Index: 2.6.18-rc6-mm1/ipc/mqueue.c
===================================================================
--- 2.6.18-rc6-mm1.orig/ipc/mqueue.c
+++ 2.6.18-rc6-mm1/ipc/mqueue.c
@@ -73,7 +73,7 @@ struct mqueue_inode_info {
struct mq_attr attr;

struct sigevent notify;
- pid_t notify_owner;
+ struct pid* notify_owner;
struct user_struct *user; /* user who created, for accounting */
struct sock *notify_sock;
struct sk_buff *notify_cookie;
@@ -134,7 +134,7 @@ static struct inode *mqueue_get_inode(st
INIT_LIST_HEAD(&info->e_wait_q[0].list);
INIT_LIST_HEAD(&info->e_wait_q[1].list);
info->messages = NULL;
- info->notify_owner = 0;
+ info->notify_owner = NULL;
info->qsize = 0;
info->user = NULL; /* set when all is ok */
memset(&info->attr, 0, sizeof(info->attr));
@@ -338,7 +338,7 @@ static ssize_t mqueue_read_file(struct f
(info->notify_owner &&
info->notify.sigev_notify == SIGEV_SIGNAL) ?
info->notify.sigev_signo : 0,
- info->notify_owner);
+ pid_nr(info->notify_owner));
spin_unlock(&info->lock);
buffer[sizeof(buffer)-1] = '\0';
slen = strlen(buffer)+1;
@@ -363,7 +363,7 @@ static int mqueue_flush_file(struct file
struct mqueue_inode_info *info = MQUEUE_I(filp->f_dentry->d_inode);

spin_lock(&info->lock);
- if (current->tgid == info->notify_owner)
+ if (task_tgid(current) == info->notify_owner)
remove_notification(info);

spin_unlock(&info->lock);
@@ -518,8 +518,8 @@ static void __do_notify(struct mqueue_in
sig_i.si_pid = current->tgid;
sig_i.si_uid = current->uid;

- kill_proc_info(info->notify.sigev_signo,
- &sig_i, info->notify_owner);
+ kill_pid_info(info->notify.sigev_signo,
+ &sig_i, info->notify_owner);
break;
case SIGEV_THREAD:
set_cookie(info->notify_cookie, NOTIFY_WOKENUP);
@@ -528,7 +528,8 @@ static void __do_notify(struct mqueue_in
break;
}
/* after notification unregisters process */
- info->notify_owner = 0;
+ put_pid(info->notify_owner);
+ info->notify_owner = NULL;
}
wake_up(&info->wait_q);
}
@@ -566,12 +567,13 @@ static long prepare_timeout(const struct

static void remove_notification(struct mqueue_inode_info *info)
{
- if (info->notify_owner != 0 &&
+ if (info->notify_owner != NULL &&
info->notify.sigev_notify == SIGEV_THREAD) {
set_cookie(info->notify_cookie, NOTIFY_REMOVED);
netlink_sendskb(info->notify_sock, info->notify_cookie, 0);
}
- info->notify_owner = 0;
+ put_pid(info->notify_owner);
+ info->notify_owner = NULL;
}

static int mq_attr_ok(struct mq_attr *attr)
@@ -1062,11 +1064,11 @@ retry:
ret = 0;
spin_lock(&info->lock);
if (u_notification == NULL) {
- if (info->notify_owner == current->tgid) {
+ if (info->notify_owner == task_tgid(current)) {
remove_notification(info);
inode->i_atime = inode->i_ctime = CURRENT_TIME;
}
- } else if (info->notify_owner != 0) {
+ } else if (info->notify_owner != NULL) {
ret = -EBUSY;
} else {
switch (notification.sigev_notify) {
@@ -1086,7 +1088,8 @@ retry:
info->notify.sigev_notify = SIGEV_SIGNAL;
break;
}
- info->notify_owner = current->tgid;
+
+ info->notify_owner = get_pid(task_tgid(current));
inode->i_atime = inode->i_ctime = CURRENT_TIME;
}
spin_unlock(&info->lock);


2006-09-09 02:39:58

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Cedric Le Goater <[email protected]> writes:

> message queues can signal a process waiting for a message.
>
> this patch replaces the pid_t value with a struct pid to avoid pid wrap
> around problems.
>
> Signed-off-by: Cedric Le Goater <[email protected]>
> Cc: Eric Biederman <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: [email protected]

Signed-off-by: Eric Biederman <[email protected]>

I was just about to send out this patch in a couple more hours.
So expect the fact we wrote the same code is a good sign :)


Eric

2006-09-11 10:17:36

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Eric W. Biederman wrote:
> Cedric Le Goater <[email protected]> writes:
>
>> message queues can signal a process waiting for a message.
>>
>> this patch replaces the pid_t value with a struct pid to avoid pid wrap
>> around problems.
>>
>> Signed-off-by: Cedric Le Goater <[email protected]>
>> Cc: Eric Biederman <[email protected]>
>> Cc: Andrew Morton <[email protected]>
>> Cc: [email protected]
>
> Signed-off-by: Eric Biederman <[email protected]>
>
> I was just about to send out this patch in a couple more hours.

Well, you did the same with the usb/devio.c and friends :)

> So expect the fact we wrote the same code is a good sign :)

How does oleg feel about it ? I've seen some long thread on possible race
conditions with put_pid() and solutions with rcu. I didn't quite get all of
it ... it will need another run for me.


On the "pid_t to struct pid*" topic:

* I started smbfs and realized it was useless.

* in the following, the init process is being killed directly using 1. I'm
not sure how useful it would be to use a struct pid. To begin with, may be
they could use a :

kill_init(int signum, int priv)

./arch/mips/sgi-ip32/ip32-reset.c
./arch/powerpc/platforms/iseries/mf.c
./drivers/parisc/power.c
./drivers/char/snsc_event.c
./kernel/sys.c
./kernel/sysctl.c
./drivers/char/nwbutton.c
./drivers/s390/s390mach.c

* some more drivers,
* some more kthread to convert


C.

2006-09-11 11:10:23

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Cedric Le Goater <[email protected]> writes:

> Eric W. Biederman wrote:
>> Cedric Le Goater <[email protected]> writes:
>>
>>> message queues can signal a process waiting for a message.
>>>
>>> this patch replaces the pid_t value with a struct pid to avoid pid wrap
>>> around problems.
>>>
>>> Signed-off-by: Cedric Le Goater <[email protected]>
>>> Cc: Eric Biederman <[email protected]>
>>> Cc: Andrew Morton <[email protected]>
>>> Cc: [email protected]
>>
>> Signed-off-by: Eric Biederman <[email protected]>
>>
>> I was just about to send out this patch in a couple more hours.
>
> Well, you did the same with the usb/devio.c and friends :)

Good. The you should be familiar enough with it to review my patch
and make certain I didn't do anything stupid :)

>> So expect the fact we wrote the same code is a good sign :)
>
> How does oleg feel about it ? I've seen some long thread on possible race
> conditions with put_pid() and solutions with rcu. I didn't quite get all of
> it ... it will need another run for me.

Short. Oleg felt it was a shame that locking was needed to use a
struct pid.

While parsing that I realized my second vt patch that deals with
vt_pid (the pid for console switching) has a subtle race, and
that patch needs to be reworked.

We confused each other. :)

> On the "pid_t to struct pid*" topic:
>
> * I started smbfs and realized it was useless.

Killing the user space part is useless?
I thought that is what I saw happening.

Of course I don't frequently mount smbfs.

> * in the following, the init process is being killed directly using 1. I'm
> not sure how useful it would be to use a struct pid. To begin with, may be
> they could use a :
>
> kill_init(int signum, int priv)

An interesting notion. The other half of them use cad_pid.
Converting that is going to need some sysctl work, so I have been
ignoring it temporarily.

Filling in a struct pid through sysctl is extremely ugly at the
moment, plus cad_pid needs some locking.

> ./arch/mips/sgi-ip32/ip32-reset.c
> ./arch/powerpc/platforms/iseries/mf.c
> ./drivers/parisc/power.c
> ./drivers/char/snsc_event.c
> ./kernel/sys.c
> ./kernel/sysctl.c
> ./drivers/char/nwbutton.c
> ./drivers/s390/s390mach.c
>
> * some more drivers,
> * some more kthread to convert

Ok. Time to exchange some status information, before I roll over and
go back to sleep.



My patch todo list (almost a series file) currently looks like:
> n_r396r
> fs3270-Change-to-use-struct-pid.txt
> smbfs-Make-conn_pid-a-struct-pid.txt
> ncpfs-Use-struct-pid-to-track-the-userspace-watchdog-process.txt
>
> Don-t-use-kill_pg-in-the-sunos-compatibility-code.txt
>
> usbatm-use-kthread-api (I think I have this one)
I did usbatm mostly to figure out why kthread conversions seem
to be so hard, and got lucky this one wasn't too ugly.

> The-dvb_core-needs-to-use-the-kthread-api-not-kernel-threads.txt
> nfs-Note-we-need-to-start-using-the-kthreads-api.txt

dvb-core I have only started looking at.
nfs I noticed it is the svc stuff that matters.

usbatm, dvb-core, and nfs are the 3 kernel_thread users
that also use kill_proc, and thus are high on my immediate hit list.

> pid-Replace-session_of_pgrp-with-pgrp_in_current_session.txt
> pid-Use-struct-pid-for-talking-about-process-groups-in-exit.c.txt
> pid-Replace-is_orphaned_pgrp-with-is_current_pgrp_orphaned.txt
>
> tty-Update-the-tty-layer-to-work-with-struct-pid.txt
I need to ensure I don't have a race with task->signal->tty_old_pgrp.
tty_old_pgrp is a weird notion that I haven't fully wrapped my head
around yet.

> pid-Remove-use-of-old-do_each_task_pid-while_each_task_pid.txt
>
> Rewrite-kill_something_info-so-it-uses-newer-helpers.txt
>
> pid-Remove-now-unused-do_each_task_pid-and-while_each_task_pid.txt
> Remove-the-now-unused-kill_pg-kill_pg_info-and-__kill_pg_info.txt
>
>
> pid-Better-tests-for-same-thread-group-membership.txt
> pid-Cleanup-the-pid-equality-tests.txt
> pid-Track-the-sending-pid-of-a-queued-signal.txt
> proc-Use-pid_nr-in-array.c-so-the-code-is-foobar-safe.txt
>
> sysctl-Implement-get_data-put_data.txt
>
> cad-pid (killing init)

Once the above list is processed none of the old none of the signal
functions that take a pid_t is needed anymore.
i.e. kill_proc, kill_pg, and do_each_task_pid will be removable.

I have at least a first draft of everything on my list except for the
kthread conversions which I just started messing with yesterday. But
don't worry about beating me to something if you feel you have it
complete. That just means I will have enough of a clue to review your
code :)

Eric

2006-09-11 14:05:58

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Eric W. Biederman wrote:

>>> I was just about to send out this patch in a couple more hours.
>> Well, you did the same with the usb/devio.c and friends :)
>
> Good. The you should be familiar enough with it to review my patch
> and make certain I didn't do anything stupid :)

well, the least i can try ...

>>> So expect the fact we wrote the same code is a good sign :)
>> How does oleg feel about it ? I've seen some long thread on possible race
>> conditions with put_pid() and solutions with rcu. I didn't quite get all of
>> it ... it will need another run for me.
>
> Short. Oleg felt it was a shame that locking was needed to use a
> struct pid.
>
> While parsing that I realized my second vt patch that deals with
> vt_pid (the pid for console switching) has a subtle race, and
> that patch needs to be reworked.
>
> We confused each other. :)
>
>> On the "pid_t to struct pid*" topic:
>>
>> * I started smbfs and realized it was useless.
>
> Killing the user space part is useless?
> I thought that is what I saw happening.

smb_fill_super() says :

if (warn_count < 5) {
warn_count++;
printk(KERN_EMERG "smbfs is deprecated and will be removed in"
" December, 2006. Please migrate to cifs\n");
}

So, i guess we should forget about it and spend our time on the cifs
kthread instead.

> Of course I don't frequently mount smbfs.
>
>> * in the following, the init process is being killed directly using 1. I'm
>> not sure how useful it would be to use a struct pid. To begin with, may be
>> they could use a :
>>
>> kill_init(int signum, int priv)
>
> An interesting notion. The other half of them use cad_pid.

yes.

> Converting that is going to need some sysctl work, so I have been
> ignoring it temporarily.
>
> Filling in a struct pid through sysctl is extremely ugly at the
> moment, plus cad_pid needs some locking.

Which distros use /proc/sys/kernel/cad_pid and why ? I can image the need
but i didn't find much on the topic.

> My patch todo list (almost a series file) currently looks like:
>> n_r396r
>> fs3270-Change-to-use-struct-pid.txt

done that. will send to martin for review.

>> smbfs-Make-conn_pid-a-struct-pid.txt

deprecated in december. so we could just forget about it.

>> ncpfs-Use-struct-pid-to-track-the-userspace-watchdog-process.txt
>>
>> Don-t-use-kill_pg-in-the-sunos-compatibility-code.txt
>>
>> usbatm-use-kthread-api (I think I have this one)
> I did usbatm mostly to figure out why kthread conversions seem
> to be so hard, and got lucky this one wasn't too ugly.

argh. i've done also and i just send my second version of the patch to the
maintainer Duncan Sands.

This one might just be useless also because greg kh has a patch in -mm to
enable multithread probing of USB devices.

>> The-dvb_core-needs-to-use-the-kthread-api-not-kernel-threads.txt
>> nfs-Note-we-need-to-start-using-the-kthreads-api.txt
>
> dvb-core I have only started looking at.

suka and i have sent patches to fix :

drivers/media/video/tvaudio.c
drivers/media/video/saa7134/saa7134-tvaudio.c

we are no waiting for the maintainer feedback.

> nfs I noticed it is the svc stuff that matters.
>
> usbatm, dvb-core, and nfs are the 3 kernel_thread users
> that also use kill_proc, and thus are high on my immediate hit list.

nfs is also full of signal_pending() ...

>> pid-Replace-session_of_pgrp-with-pgrp_in_current_session.txt
>> pid-Use-struct-pid-for-talking-about-process-groups-in-exit.c.txt
>> pid-Replace-is_orphaned_pgrp-with-is_current_pgrp_orphaned.txt
>>
>> tty-Update-the-tty-layer-to-work-with-struct-pid.txt
> I need to ensure I don't have a race with task->signal->tty_old_pgrp.
> tty_old_pgrp is a weird notion that I haven't fully wrapped my head
> around yet.
>
>> pid-Remove-use-of-old-do_each_task_pid-while_each_task_pid.txt
>>
>> Rewrite-kill_something_info-so-it-uses-newer-helpers.txt
>>
>> pid-Remove-now-unused-do_each_task_pid-and-while_each_task_pid.txt
>> Remove-the-now-unused-kill_pg-kill_pg_info-and-__kill_pg_info.txt
>>
>>
>> pid-Better-tests-for-same-thread-group-membership.txt
>> pid-Cleanup-the-pid-equality-tests.txt
>> pid-Track-the-sending-pid-of-a-queued-signal.txt

is that about updating the siginfos in collect_signal() to hold the right
pid value depending on the pid namespace they are being received ?

>> proc-Use-pid_nr-in-array.c-so-the-code-is-foobar-safe.txt
>>
>> sysctl-Implement-get_data-put_data.txt
>>
>> cad-pid (killing init)
>
> Once the above list is processed none of the old none of the signal
> functions that take a pid_t is needed anymore.
> i.e. kill_proc, kill_pg, and do_each_task_pid will be removable.
>
> I have at least a first draft of everything on my list except for the
> kthread conversions which I just started messing with yesterday. But
> don't worry about beating me to something if you feel you have it
> complete. That just means I will have enough of a clue to review your
> code :)

good list ! I look at it in details.

thanks,

C.

2006-09-11 15:48:42

by Oleg Nesterov

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

On 09/11, Cedric Le Goater wrote:
>
> Eric W. Biederman wrote:
> > Cedric Le Goater <[email protected]> writes:
> >
> >> message queues can signal a process waiting for a message.
> >>
> >> this patch replaces the pid_t value with a struct pid to avoid pid wrap
> >> around problems.
> >>
> >> Signed-off-by: Cedric Le Goater <[email protected]>
> >> Cc: Eric Biederman <[email protected]>
> >> Cc: Andrew Morton <[email protected]>
> >> Cc: [email protected]
> >
> > Signed-off-by: Eric Biederman <[email protected]>
> >
> > I was just about to send out this patch in a couple more hours.
>
> Well, you did the same with the usb/devio.c and friends :)
>
> > So expect the fact we wrote the same code is a good sign :)
>
> How does oleg feel about it ? I've seen some long thread on possible race
> conditions with put_pid() and solutions with rcu. I didn't quite get all of
> it ... it will need another run for me.

I assume you are talking about this patch:
http://marc.theaimsgroup.com/?l=linux-mm-commits&m=115773820415171

I think it's ok, info->notify_owner is always used under info->lock.

This is simple. If, for example, mqueue_read_file() didn't take info->lock,
then we have a problem: pid_nr() may read a freed memory in case when
__do_notify()->put_pid() happens at the same time.

In this context info->notify_owner is a usual refcounted object, no special
attention is needed.

Oleg.

2006-09-11 19:02:22

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid


Cedric you mentioned a couple of other patches that are in flight.
In the future could you please Cc: the containers list so independent
efforts are less likely to duplicate work, and we are more likely
to review each others patches instead?

Cedric Le Goater <[email protected]> writes:

> Eric W. Biederman wrote:
>
>>>> I was just about to send out this patch in a couple more hours.
>>> Well, you did the same with the usb/devio.c and friends :)
>>
>> Good. The you should be familiar enough with it to review my patch
>> and make certain I didn't do anything stupid :)
>
> well, the least i can try ...
>

>>> * I started smbfs and realized it was useless.
>>
>> Killing the user space part is useless?
>> I thought that is what I saw happening.
>
> smb_fill_super() says :
>
> if (warn_count < 5) {
> warn_count++;
> printk(KERN_EMERG "smbfs is deprecated and will be removed in"
> " December, 2006. Please migrate to cifs\n");
> }
>
> So, i guess we should forget about it and spend our time on the cifs
> kthread instead.

Sure. Although in this instance the changes are simple enough I will probably
send the patch anyway :) That at least explains why you figured it was
useless work.


>> Of course I don't frequently mount smbfs.
>>
>>> * in the following, the init process is being killed directly using 1. I'm
>>> not sure how useful it would be to use a struct pid. To begin with, may be
>>> they could use a :
>>>
>>> kill_init(int signum, int priv)
>>
>> An interesting notion. The other half of them use cad_pid.
>
> yes.
>
>> Converting that is going to need some sysctl work, so I have been
>> ignoring it temporarily.
>>
>> Filling in a struct pid through sysctl is extremely ugly at the
>> moment, plus cad_pid needs some locking.
>
> Which distros use /proc/sys/kernel/cad_pid and why ? I can image the need
> but i didn't find much on the topic.

I'm not at all certain, and I'm not even certain I care. The concept
is there in the code so it needs to be dealt with. Although if I we
extend the cad_pid concept it may make a difference.


>> My patch todo list (almost a series file) currently looks like:
>>> n_r396r
>>> fs3270-Change-to-use-struct-pid.txt
>
> done that. will send to martin for review.

Added to my queue of pending patches to look at review.


>>> ncpfs-Use-struct-pid-to-track-the-userspace-watchdog-process.txt
>>>
>>> Don-t-use-kill_pg-in-the-sunos-compatibility-code.txt
>>>
>>> usbatm-use-kthread-api (I think I have this one)
>> I did usbatm mostly to figure out why kthread conversions seem
>> to be so hard, and got lucky this one wasn't too ugly.
>
> argh. i've done also and i just send my second version of the patch to the
> maintainer Duncan Sands.
>
> This one might just be useless also because greg kh has a patch in -mm to
> enable multithread probing of USB devices.


Added to my queue of pending patches to track down and reivew.


>>> The-dvb_core-needs-to-use-the-kthread-api-not-kernel-threads.txt
>>> nfs-Note-we-need-to-start-using-the-kthreads-api.txt
>>
>> dvb-core I have only started looking at.
>
> suka and i have sent patches to fix :
>
> drivers/media/video/tvaudio.c
> drivers/media/video/saa7134/saa7134-tvaudio.c
>
> we are no waiting for the maintainer feedback.

Ok. I think I saw a little of that.

>> nfs I noticed it is the svc stuff that matters.
>>
>> usbatm, dvb-core, and nfs are the 3 kernel_thread users
>> that also use kill_proc, and thus are high on my immediate hit list.
>
> nfs is also full of signal_pending() ...

Yes, there is a lot to read and understand before I can confidently
do much with nfs.

>>> pid-Better-tests-for-same-thread-group-membership.txt
>>> pid-Cleanup-the-pid-equality-tests.txt
>>> pid-Track-the-sending-pid-of-a-queued-signal.txt
>
> is that about updating the siginfos in collect_signal() to hold the right
> pid value depending on the pid namespace they are being received ?

Yes in send_signal, and in collect signal. To make it work easily I needed
to add a struct pid to struct sigqueue. So in send_signal I generate
the struct pid from the pid_t value and in collect signal I regenerate
the numeric value.


Eric

2006-09-11 21:53:44

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Eric W. Biederman wrote:

> Cedric you mentioned a couple of other patches that are in flight.
> In the future could you please Cc: the containers list so independent
> efforts are less likely to duplicate work, and we are more likely
> to review each others patches instead?

yes sure, i was relying on the openvz wiki to avoid duplicated efforts on
this topic but i guess email is just the one and only tool for this kind of
development :)

[ ... ]

>>> Converting that is going to need some sysctl work, so I have been
>>> ignoring it temporarily.
>>>
>>> Filling in a struct pid through sysctl is extremely ugly at the
>>> moment, plus cad_pid needs some locking.
>> Which distros use /proc/sys/kernel/cad_pid and why ? I can image the need
>> but i didn't find much on the topic.
>
> I'm not at all certain, and I'm not even certain I care. The concept
> is there in the code so it needs to be dealt with.

OK. It would be nice to make sure this is still in use before trying to
deal with /proc/sys/kernel/cad_pid.

> Although if I we extend the cad_pid concept it may make a difference.

what do you mean by extending cad_pid ? kill_init() ?

[ ... ]

>> is that about updating the siginfos in collect_signal() to hold the right
>> pid value depending on the pid namespace they are being received ?
>
> Yes in send_signal, and in collect signal. To make it work easily I needed
> to add a struct pid to struct sigqueue. So in send_signal I generate
> the struct pid from the pid_t value and in collect signal I regenerate
> the numeric value.

OK. That's what i imagined also but we need a bit more of the pid namespace
to regenerate the numerical value. So, how will you convert this 'struct
pid*' in a pid value using the current pid namespace ?

thinking aloud :

* if the pid namespace of the sending struct pid and current match,
use nr.
* if they don't,
if the sending pid namespace is the ancestor of the current pid
namespace
use 0
else
it's a bug.

struct pid* needs a pid namespace attribute and pid namespace needs to know
its parent.

C.

2006-09-12 01:23:16

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Cedric Le Goater <[email protected]> writes:

> Eric W. Biederman wrote:
>
>> Cedric you mentioned a couple of other patches that are in flight.
>> In the future could you please Cc: the containers list so independent
>> efforts are less likely to duplicate work, and we are more likely
>> to review each others patches instead?
>
> yes sure, i was relying on the openvz wiki to avoid duplicated efforts on
> this topic but i guess email is just the one and only tool for this kind of
> development :)

Sure. Especially when it comes to helping review each others code :)
Not duplicating work is not really my goal, not submitting a patch
after a patch has been reviewed and accepted is.

Plus we need patch review.

Several people working on a patch in parallel if it is difficult
can frequently find a solution that a single person would miss.

>>>> Filling in a struct pid through sysctl is extremely ugly at the
>>>> moment, plus cad_pid needs some locking.
>>> Which distros use /proc/sys/kernel/cad_pid and why ? I can image the need
>>> but i didn't find much on the topic.
>>
>> I'm not at all certain, and I'm not even certain I care. The concept
>> is there in the code so it needs to be dealt with.
>
> OK. It would be nice to make sure this is still in use before trying to
> deal with /proc/sys/kernel/cad_pid.
>
>> Although if I we extend the cad_pid concept it may make a difference.
>
> what do you mean by extending cad_pid ? kill_init() ?

My meaning was every time we are sending a signal to init. It is quite
possible we should be using cad_pid instead.

>>> is that about updating the siginfos in collect_signal() to hold the right
>>> pid value depending on the pid namespace they are being received ?
>>
>> Yes in send_signal, and in collect signal. To make it work easily I needed
>> to add a struct pid to struct sigqueue. So in send_signal I generate
>> the struct pid from the pid_t value and in collect signal I regenerate
>> the numeric value.
>
> OK. That's what i imagined also but we need a bit more of the pid namespace
> to regenerate the numerical value. So, how will you convert this 'struct
> pid*' in a pid value using the current pid namespace ?

By calling pid_nr :) The question I guess is how will pid_nr be implemented.

> thinking aloud :
>
> * if the pid namespace of the sending struct pid and current match,
> use nr.
> * if they don't,
> if the sending pid namespace is the ancestor of the current pid
> namespace
> use 0
> else
> it's a bug.
>
> struct pid* needs a pid namespace attribute and pid namespace needs to know
> its parent.

Yes, that sounds correct.

There is also the case that should not come up with signals where
we have a pid from a child namespace, that we should also be able to
compute the pid for.

In essence I intend to have a list of pid_namespace, pid_t pairs connected
to a struct pid that we can look through to find the appropriate pid.

Eric


2006-09-12 11:06:01

by Herbert Poetzl

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

On Mon, Sep 11, 2006 at 01:01:18PM -0600, Eric W. Biederman wrote:
>
> Cedric you mentioned a couple of other patches that are in flight.
> In the future could you please Cc: the containers list so independent
> efforts are less likely to duplicate work, and we are more likely
> to review each others patches instead?
>
> Cedric Le Goater <[email protected]> writes:
>
> > Eric W. Biederman wrote:
> >
> >>>> I was just about to send out this patch in a couple more hours.
> >>> Well, you did the same with the usb/devio.c and friends :)
> >>
> >> Good. The you should be familiar enough with it to review my patch
> >> and make certain I didn't do anything stupid :)
> >
> > well, the least i can try ...
> >
>
> >>> * I started smbfs and realized it was useless.
> >>
> >> Killing the user space part is useless?
> >> I thought that is what I saw happening.
> >
> > smb_fill_super() says :
> >
> > if (warn_count < 5) {
> > warn_count++;
> > printk(KERN_EMERG "smbfs is deprecated and will be removed in"
> > " December, 2006. Please migrate to cifs\n");
> > }
> >
> > So, i guess we should forget about it and spend our time on the cifs
> > kthread instead.
>
> Sure. Although in this instance the changes are simple enough I will
> probably send the patch anyway :) That at least explains why you
> figured it was useless work.
>
>
> >> Of course I don't frequently mount smbfs.
> >>
> >>> * in the following, the init process is being killed directly
> >>> using 1. I'm not sure how useful it would be to use a struct pid.
> >>> To begin with, may be they could use a :
> >>>
> >>> kill_init(int signum, int priv)
> >>
> >> An interesting notion. The other half of them use cad_pid.
> >
> > yes.
> >
> >> Converting that is going to need some sysctl work, so I have been
> >> ignoring it temporarily.
> >>
> >> Filling in a struct pid through sysctl is extremely ugly at the
> >> moment, plus cad_pid needs some locking.
> >
> > Which distros use /proc/sys/kernel/cad_pid and why ? I can image the
> > need but i didn't find much on the topic.
>
> I'm not at all certain, and I'm not even certain I care. The concept
> is there in the code so it needs to be dealt with. Although if I we
> extend the cad_pid concept it may make a difference.
>
> >> My patch todo list (almost a series file) currently looks like:
> >>
> >>> n_r396r fs3270-Change-to-use-struct-pid.txt
> >
> > done that. will send to martin for review.
>
> Added to my queue of pending patches to look at review.
>
> >>> ncpfs-Use-struct-pid-to-track-the-userspace-watchdog-process.txt
> >>>
> >>> Don-t-use-kill_pg-in-the-sunos-compatibility-code.txt
> >>>
> >>> usbatm-use-kthread-api (I think I have this one)
> >>
> >> I did usbatm mostly to figure out why kthread conversions seem to
> >> be so hard, and got lucky this one wasn't too ugly.
> >
> > argh. i've done also and i just send my second version of the patch
> > to the maintainer Duncan Sands.
> >
> > This one might just be useless also because greg kh has a patch in
> > -mm to enable multithread probing of USB devices.
>
> Added to my queue of pending patches to track down and reivew.
>
>
> >>> The-dvb_core-needs-to-use-the-kthread-api-not-kernel-threads.txt
> >>> nfs-Note-we-need-to-start-using-the-kthreads-api.txt
> >>
> >> dvb-core I have only started looking at.
> >
> > suka and i have sent patches to fix :
> >
> > drivers/media/video/tvaudio.c
> > drivers/media/video/saa7134/saa7134-tvaudio.c
> >
> > we are no waiting for the maintainer feedback.
>
> Ok. I think I saw a little of that.
>
> >> nfs I noticed it is the svc stuff that matters.
> >>
> >> usbatm, dvb-core, and nfs are the 3 kernel_thread users
> >> that also use kill_proc, and thus are high on my immediate hit list.

> > nfs is also full of signal_pending() ...

> Yes, there is a lot to read and understand before I can confidently
> do much with nfs.

I already did a lot of adjustments to the nfs system, and
I poked aound in dvb-core before, so I will take a look
at this in the next few days, at least the switch to the
kthread api should not be a big deal ...

HTH,
Herbert

> >>> pid-Better-tests-for-same-thread-group-membership.txt
> >>> pid-Cleanup-the-pid-equality-tests.txt
> >>> pid-Track-the-sending-pid-of-a-queued-signal.txt
> >
> > is that about updating the siginfos in collect_signal() to hold the right
> > pid value depending on the pid namespace they are being received ?
>
> Yes in send_signal, and in collect signal. To make it work easily I
> needed to add a struct pid to struct sigqueue. So in send_signal I
> generate the struct pid from the pid_t value and in collect signal I
> regenerate the numeric value.
>
>
> Eric
>
> _______________________________________________
> Containers mailing list
> [email protected]
> https://lists.osdl.org/mailman/listinfo/containers

2006-09-12 15:15:05

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Herbert Poetzl <[email protected]> writes:

> I already did a lot of adjustments to the nfs system, and
> I poked aound in dvb-core before, so I will take a look
> at this in the next few days, at least the switch to the
> kthread api should not be a big deal ...

Ok. If you can get this it would be great.

To some extent the last holdouts on the kernel_thread api seem to
be the ones that are not trivial to convert :(

Eric

2006-09-12 15:37:36

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Eric W. Biederman wrote:

[ ... ]

> There is also the case that should not come up with signals where
> we have a pid from a child namespace, that we should also be able to
> compute the pid for.

I don't understand how a signal can come from a child pid namespace ?

> In essence I intend to have a list of pid_namespace, pid_t pairs connected
> to a struct pid that we can look through to find the appropriate pid.

yes, that's the purpose of pid_nr() I guess.

This list would contain in nearly all cases a single pair (current pid
namespace, pid value). It will contain 2 pairs for a task that has unshared
its pid namespace : a pair for the current pid namespace, that needs to
allocated when unshare() is called, and one pair for the ancestor pid
namespace which is already allocated.

Do you see more ?

C.

2006-09-12 16:04:36

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [patch -mm] update mq_notify to use a struct pid

Cedric Le Goater <[email protected]> writes:

> Eric W. Biederman wrote:
>
> [ ... ]
>
>> There is also the case that should not come up with signals where
>> we have a pid from a child namespace, that we should also be able to
>> compute the pid for.
>
> I don't understand how a signal can come from a child pid namespace ?

SIG_CHLD is the only case where I think we will be sending a signal
from the child pid namespace.

Reading pids from the status files in /proc, from a parent namespace,
is another example where we need to deal with the pid of children.

>> In essence I intend to have a list of pid_namespace, pid_t pairs connected
>> to a struct pid that we can look through to find the appropriate pid.
>
> yes, that's the purpose of pid_nr() I guess.
>
> This list would contain in nearly all cases a single pair (current pid
> namespace, pid value). It will contain 2 pairs for a task that has unshared
> its pid namespace : a pair for the current pid namespace, that needs to
> allocated when unshare() is called, and one pair for the ancestor pid
> namespace which is already allocated.
>
> Do you see more ?

I don't see the list getting longer until we get into a nested pid namespaces.

As long as the interface is well defined for the container in a container
case I don't mind having additional restrictions.

I will note that you can get some extremely weird interactions if
you do things like open a file descriptor in the parent pid namespace.
Fork two children each child in a different pid_namespaces.
fcntl(F_SETOWN) is called in one child, and fcntl(F_GETOWN) is called
in the other child.

So we can't just call BUG_ON, if we have can't find the namespace.
But returning 0 from pid_nr should be fine.

Eric

2006-09-14 20:01:05

by Herbert Poetzl

[permalink] [raw]
Subject: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110


Okay, as I promised, I had a first shot at the
dvb kernel_thread to kthread API port, and here
is the result, which is running fine here since
yesterday, including module load/unload and
software suspend (which doesn't work as expected
with or without this patch :), I didn't convert
the dvb_ca_en50221 as I do not have such an
interface, but if the conversion process is fine
with the v4l-dvb maintainers, it should not be
a problem to send a patch for that too ...

best,
Herbert

Signed-off-by: Herbert Poetzl <[email protected]>

diff -NurpP linux-2.6.18-rc6/drivers/media/dvb/dvb-core/dvb_frontend.c linux-2.6.18-rc6-kthread.v02.3/drivers/media/dvb/dvb-core/dvb_frontend.c
--- linux-2.6.18-rc6/drivers/media/dvb/dvb-core/dvb_frontend.c 2006-09-12 18:16:12 +0200
+++ linux-2.6.18-rc6-kthread.v02.3/drivers/media/dvb/dvb-core/dvb_frontend.c 2006-09-14 21:23:37 +0200
@@ -36,6 +36,7 @@
#include <linux/list.h>
#include <linux/suspend.h>
#include <linux/jiffies.h>
+#include <linux/kthread.h>
#include <asm/processor.h>

#include "dvb_frontend.h"
@@ -100,7 +101,7 @@ struct dvb_frontend_private {
struct semaphore sem;
struct list_head list_head;
wait_queue_head_t wait_queue;
- pid_t thread_pid;
+ struct task_struct *thread;
unsigned long release_jiffies;
unsigned int exit;
unsigned int wakeup;
@@ -508,19 +509,11 @@ static int dvb_frontend_thread(void *dat
struct dvb_frontend *fe = data;
struct dvb_frontend_private *fepriv = fe->frontend_priv;
unsigned long timeout;
- char name [15];
fe_status_t s;
struct dvb_frontend_parameters *params;

dprintk("%s\n", __FUNCTION__);

- snprintf (name, sizeof(name), "kdvb-fe-%i", fe->dvb->num);
-
- lock_kernel();
- daemonize(name);
- sigfillset(&current->blocked);
- unlock_kernel();
-
fepriv->check_wrapped = 0;
fepriv->quality = 0;
fepriv->delay = 3*HZ;
@@ -534,14 +527,16 @@ static int dvb_frontend_thread(void *dat
up(&fepriv->sem); /* is locked when we enter the thread... */

timeout = wait_event_interruptible_timeout(fepriv->wait_queue,
- dvb_frontend_should_wakeup(fe),
- fepriv->delay);
- if (0 != dvb_frontend_is_exiting(fe)) {
+ dvb_frontend_should_wakeup(fe) || kthread_should_stop(),
+ fepriv->delay);
+
+ if (kthread_should_stop() || dvb_frontend_is_exiting(fe)) {
/* got signal or quitting */
break;
}

- try_to_freeze();
+ if (try_to_freeze())
+ continue;

if (down_interruptible(&fepriv->sem))
break;
@@ -591,7 +586,7 @@ static int dvb_frontend_thread(void *dat
fe->ops.sleep(fe);
}

- fepriv->thread_pid = 0;
+ fepriv->thread = NULL;
mb();

dvb_frontend_wakeup(fe);
@@ -600,7 +595,6 @@ static int dvb_frontend_thread(void *dat

static void dvb_frontend_stop(struct dvb_frontend *fe)
{
- unsigned long ret;
struct dvb_frontend_private *fepriv = fe->frontend_priv;

dprintk ("%s\n", __FUNCTION__);
@@ -608,33 +602,17 @@ static void dvb_frontend_stop(struct dvb
fepriv->exit = 1;
mb();

- if (!fepriv->thread_pid)
- return;
-
- /* check if the thread is really alive */
- if (kill_proc(fepriv->thread_pid, 0, 1) == -ESRCH) {
- printk("dvb_frontend_stop: thread PID %d already died\n",
- fepriv->thread_pid);
- /* make sure the mutex was not held by the thread */
- init_MUTEX (&fepriv->sem);
+ if (!fepriv->thread)
return;
- }
-
- /* wake up the frontend thread, so it notices that fe->exit == 1 */
- dvb_frontend_wakeup(fe);

- /* wait until the frontend thread has exited */
- ret = wait_event_interruptible(fepriv->wait_queue,0 == fepriv->thread_pid);
- if (-ERESTARTSYS != ret) {
- fepriv->state = FESTATE_IDLE;
- return;
- }
+ kthread_stop(fepriv->thread);
+ init_MUTEX (&fepriv->sem);
fepriv->state = FESTATE_IDLE;

/* paranoia check in case a signal arrived */
- if (fepriv->thread_pid)
- printk("dvb_frontend_stop: warning: thread PID %d won't exit\n",
- fepriv->thread_pid);
+ if (fepriv->thread)
+ printk("dvb_frontend_stop: warning: thread %p won't exit\n",
+ fepriv->thread);
}

s32 timeval_usec_diff(struct timeval lasttime, struct timeval curtime)
@@ -684,10 +662,11 @@ static int dvb_frontend_start(struct dvb
{
int ret;
struct dvb_frontend_private *fepriv = fe->frontend_priv;
+ struct task_struct *fe_thread;

dprintk ("%s\n", __FUNCTION__);

- if (fepriv->thread_pid) {
+ if (fepriv->thread) {
if (!fepriv->exit)
return 0;
else
@@ -701,18 +680,18 @@ static int dvb_frontend_start(struct dvb

fepriv->state = FESTATE_IDLE;
fepriv->exit = 0;
- fepriv->thread_pid = 0;
+ fepriv->thread = NULL;
mb();

- ret = kernel_thread (dvb_frontend_thread, fe, 0);
-
- if (ret < 0) {
- printk("dvb_frontend_start: failed to start kernel_thread (%d)\n", ret);
+ fe_thread = kthread_run(dvb_frontend_thread, fe,
+ "kdvb-fe-%i", fe->dvb->num);
+ if (IS_ERR(fe_thread)) {
+ ret = PTR_ERR(fe_thread);
+ printk("dvb_frontend_start: failed to start kthread (%d)\n", ret);
up(&fepriv->sem);
return ret;
}
- fepriv->thread_pid = ret;
-
+ fepriv->thread = fe_thread;
return 0;
}

diff -NurpP linux-2.6.18-rc6/drivers/media/dvb/ttpci/av7110.c linux-2.6.18-rc6-kthread.v02.3/drivers/media/dvb/ttpci/av7110.c
--- linux-2.6.18-rc6/drivers/media/dvb/ttpci/av7110.c 2006-09-12 18:16:13 +0200
+++ linux-2.6.18-rc6-kthread.v02.3/drivers/media/dvb/ttpci/av7110.c 2006-09-14 21:21:03 +0200
@@ -51,6 +51,7 @@
#include <linux/firmware.h>
#include <linux/crc32.h>
#include <linux/i2c.h>
+#include <linux/kthread.h>

#include <asm/system.h>

@@ -223,11 +224,10 @@ static void recover_arm(struct av7110 *a

static void av7110_arm_sync(struct av7110 *av7110)
{
- av7110->arm_rmmod = 1;
- wake_up_interruptible(&av7110->arm_wait);
+ if (av7110->arm_thread)
+ kthread_stop(av7110->arm_thread);

- while (av7110->arm_thread)
- msleep(1);
+ av7110->arm_thread = NULL;
}

static int arm_thread(void *data)
@@ -238,17 +238,11 @@ static int arm_thread(void *data)

dprintk(4, "%p\n",av7110);

- lock_kernel();
- daemonize("arm_mon");
- sigfillset(&current->blocked);
- unlock_kernel();
-
- av7110->arm_thread = current;
-
for (;;) {
timeout = wait_event_interruptible_timeout(av7110->arm_wait,
- av7110->arm_rmmod, 5 * HZ);
- if (-ERESTARTSYS == timeout || av7110->arm_rmmod) {
+ kthread_should_stop(), 5 * HZ);
+
+ if (-ERESTARTSYS == timeout || kthread_should_stop()) {
/* got signal or told to quit*/
break;
}
@@ -276,7 +270,6 @@ static int arm_thread(void *data)
av7110->arm_errors = 0;
}

- av7110->arm_thread = NULL;
return 0;
}

@@ -2334,6 +2327,7 @@ static int __devinit av7110_attach(struc
const int length = TS_WIDTH * TS_HEIGHT;
struct pci_dev *pdev = dev->pci;
struct av7110 *av7110;
+ struct task_struct *thread;
int ret, count = 0;

dprintk(4, "dev: %p\n", dev);
@@ -2618,9 +2612,12 @@ static int __devinit av7110_attach(struc
printk ("dvb-ttpci: Warning, firmware version 0x%04x is too old. "
"System might be unstable!\n", FW_VERSION(av7110->arm_app));

- ret = kernel_thread(arm_thread, (void *) av7110, 0);
- if (ret < 0)
+ thread = kthread_run(arm_thread, (void *) av7110, "arm_mon");
+ if (IS_ERR(thread)) {
+ ret = PTR_ERR(thread);
goto err_stop_arm_9;
+ }
+ av7110->arm_thread = thread;

/* set initial volume in mixer struct */
av7110->mixer.volume_left = volume;
diff -NurpP linux-2.6.18-rc6/drivers/media/dvb/ttpci/av7110.h linux-2.6.18-rc6-kthread.v02.3/drivers/media/dvb/ttpci/av7110.h
--- linux-2.6.18-rc6/drivers/media/dvb/ttpci/av7110.h 2006-09-12 18:16:13 +0200
+++ linux-2.6.18-rc6-kthread.v02.3/drivers/media/dvb/ttpci/av7110.h 2006-09-14 21:21:03 +0200
@@ -205,7 +205,6 @@ struct av7110 {
struct task_struct *arm_thread;
wait_queue_head_t arm_wait;
u16 arm_loops;
- int arm_rmmod;

void *debi_virt;
dma_addr_t debi_bus;

2006-09-14 21:07:57

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

Herbert Poetzl wrote:
> Okay, as I promised, I had a first shot at the
> dvb kernel_thread to kthread API port, and here
> is the result, which is running fine here since
> yesterday, including module load/unload and
> software suspend (which doesn't work as expected
> with or without this patch :),

So you have such an hardware ?

[ ... ]

> @@ -600,7 +595,6 @@ static int dvb_frontend_thread(void *dat
>
> static void dvb_frontend_stop(struct dvb_frontend *fe)
> {
> - unsigned long ret;
> struct dvb_frontend_private *fepriv = fe->frontend_priv;
>
> dprintk ("%s\n", __FUNCTION__);
> @@ -608,33 +602,17 @@ static void dvb_frontend_stop(struct dvb
> fepriv->exit = 1;

do we still need the ->exit flag now that we are using kthread_stop() ?
same question for ->wakeup ?

> mb();
>
> - if (!fepriv->thread_pid)
> - return;
> -
> - /* check if the thread is really alive */
> - if (kill_proc(fepriv->thread_pid, 0, 1) == -ESRCH) {
> - printk("dvb_frontend_stop: thread PID %d already died\n",
> - fepriv->thread_pid);
> - /* make sure the mutex was not held by the thread */
> - init_MUTEX (&fepriv->sem);
> + if (!fepriv->thread)
> return;
> - }
> -
> - /* wake up the frontend thread, so it notices that fe->exit == 1 */
> - dvb_frontend_wakeup(fe);
>
> - /* wait until the frontend thread has exited */
> - ret = wait_event_interruptible(fepriv->wait_queue,0 == fepriv->thread_pid);
> - if (-ERESTARTSYS != ret) {
> - fepriv->state = FESTATE_IDLE;
> - return;
> - }
> + kthread_stop(fepriv->thread);
> + init_MUTEX (&fepriv->sem);

the use of the semaphore to synchronise the thread is complex. It will
require extra care to avoid deadlocks.

> fepriv->state = FESTATE_IDLE;
>
> /* paranoia check in case a signal arrived */
> - if (fepriv->thread_pid)
> - printk("dvb_frontend_stop: warning: thread PID %d won't exit\n",
> - fepriv->thread_pid);
> + if (fepriv->thread)
> + printk("dvb_frontend_stop: warning: thread %p won't exit\n",
> + fepriv->thread);

kthread_stop uses a completion already. so the above is real paranoia :)

> }
>
> s32 timeval_usec_diff(struct timeval lasttime, struct timeval curtime)
> @@ -684,10 +662,11 @@ static int dvb_frontend_start(struct dvb
> {
> int ret;
> struct dvb_frontend_private *fepriv = fe->frontend_priv;
> + struct task_struct *fe_thread;
>
> dprintk ("%s\n", __FUNCTION__);
>
> - if (fepriv->thread_pid) {
> + if (fepriv->thread) {
> if (!fepriv->exit)
> return 0;
> else
> @@ -701,18 +680,18 @@ static int dvb_frontend_start(struct dvb
>
> fepriv->state = FESTATE_IDLE;
> fepriv->exit = 0;
> - fepriv->thread_pid = 0;
> + fepriv->thread = NULL;
> mb();
>
> - ret = kernel_thread (dvb_frontend_thread, fe, 0);
> -
> - if (ret < 0) {
> - printk("dvb_frontend_start: failed to start kernel_thread (%d)\n", ret);
> + fe_thread = kthread_run(dvb_frontend_thread, fe,
> + "kdvb-fe-%i", fe->dvb->num);
> + if (IS_ERR(fe_thread)) {
> + ret = PTR_ERR(fe_thread);

ret could be local.


[ ... ]

2006-09-14 22:10:26

by Herbert Poetzl

[permalink] [raw]
Subject: Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

On Thu, Sep 14, 2006 at 11:07:49PM +0200, Cedric Le Goater wrote:
> Herbert Poetzl wrote:
> > Okay, as I promised, I had a first shot at the
> > dvb kernel_thread to kthread API port, and here
> > is the result, which is running fine here since
> > yesterday, including module load/unload and
> > software suspend (which doesn't work as expected
> > with or without this patch :),
>
> So you have such an hardware ?

yes, I do .. that's how I tested it :)

> [ ... ]
>
> > @@ -600,7 +595,6 @@ static int dvb_frontend_thread(void *dat
> >
> > static void dvb_frontend_stop(struct dvb_frontend *fe)
> > {
> > - unsigned long ret;
> > struct dvb_frontend_private *fepriv = fe->frontend_priv;
> >
> > dprintk ("%s\n", __FUNCTION__);
> > @@ -608,33 +602,17 @@ static void dvb_frontend_stop(struct dvb
> > fepriv->exit = 1;
>
> do we still need the ->exit flag now that we are using kthread_stop() ?
> same question for ->wakeup ?

probably not, but I didn't want to change too
much on the first try, especially I'd appreciate
some feedback to this from the maintainer(s)

> > mb();
> >
> > - if (!fepriv->thread_pid)
> > - return;
> > -
> > - /* check if the thread is really alive */
> > - if (kill_proc(fepriv->thread_pid, 0, 1) == -ESRCH) {
> > - printk("dvb_frontend_stop: thread PID %d already died\n",
> > - fepriv->thread_pid);
> > - /* make sure the mutex was not held by the thread */
> > - init_MUTEX (&fepriv->sem);
> > + if (!fepriv->thread)
> > return;
> > - }
> > -
> > - /* wake up the frontend thread, so it notices that fe->exit == 1 */
> > - dvb_frontend_wakeup(fe);
> >
> > - /* wait until the frontend thread has exited */
> > - ret = wait_event_interruptible(fepriv->wait_queue,0 == fepriv->thread_pid);
> > - if (-ERESTARTSYS != ret) {
> > - fepriv->state = FESTATE_IDLE;
> > - return;
> > - }
> > + kthread_stop(fepriv->thread);
> > + init_MUTEX (&fepriv->sem);
>
> the use of the semaphore to synchronise the thread is complex. It will
> require extra care to avoid deadlocks.

well, it 'works' quite fine for now, but yeah, I
thought about completely removing the additional
synchronization and 'jsut' go with the kthread
one, if that is sufficient ...

> > fepriv->state = FESTATE_IDLE;
> >
> > /* paranoia check in case a signal arrived */
> > - if (fepriv->thread_pid)
> > - printk("dvb_frontend_stop: warning: thread PID %d won't exit\n",
> > - fepriv->thread_pid);
> > + if (fepriv->thread)
> > + printk("dvb_frontend_stop: warning: thread %p won't exit\n",
> > + fepriv->thread);
>
> kthread_stop uses a completion already. so the above is real paranoia :)

again, I think this will go away soon :)

> > }
> >
> > s32 timeval_usec_diff(struct timeval lasttime, struct timeval curtime)
> > @@ -684,10 +662,11 @@ static int dvb_frontend_start(struct dvb
> > {
> > int ret;
> > struct dvb_frontend_private *fepriv = fe->frontend_priv;
> > + struct task_struct *fe_thread;
> >
> > dprintk ("%s\n", __FUNCTION__);
> >
> > - if (fepriv->thread_pid) {
> > + if (fepriv->thread) {
> > if (!fepriv->exit)
> > return 0;
> > else
> > @@ -701,18 +680,18 @@ static int dvb_frontend_start(struct dvb
> >
> > fepriv->state = FESTATE_IDLE;
> > fepriv->exit = 0;
> > - fepriv->thread_pid = 0;
> > + fepriv->thread = NULL;
> > mb();
> >
> > - ret = kernel_thread (dvb_frontend_thread, fe, 0);
> > -
> > - if (ret < 0) {
> > - printk("dvb_frontend_start: failed to start kernel_thread (%d)\n", ret);
> > + fe_thread = kthread_run(dvb_frontend_thread, fe,
> > + "kdvb-fe-%i", fe->dvb->num);
> > + if (IS_ERR(fe_thread)) {
> > + ret = PTR_ERR(fe_thread);
>
> ret could be local.

correct, will fix that up in the next round

thanks for the feedback,
Herbert

> [ ... ]
>
> _______________________________________________
> Containers mailing list
> [email protected]
> https://lists.osdl.org/mailman/listinfo/containers

2006-11-17 01:50:08

by Andrew de Quincey

[permalink] [raw]
Subject: Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

[snip]

> correct, will fix that up in the next round
>
> thanks for the feedback,
> Herbert

Hi - the conversion looks good to me.. I can't really offer any more
constructive suggestions beyond what Cedric has already said.

Theres another thread in dvb_ca_en50221.c that could be converted as well
though, hint hint ;)

Apologies for the delay in this reply - I've been hibernating for a bit.

2006-11-22 14:56:24

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [Devel] Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

Andrew de Quincey wrote:

> Hi - the conversion looks good to me.. I can't really offer any more
> constructive suggestions beyond what Cedric has already said.

ok. so, should we just resend a refreshed version of the patch when 2.6.19
comes out ?

> Theres another thread in dvb_ca_en50221.c that could be converted as well
> though, hint hint ;)

ok ok :) i'll look at it ...

> Apologies for the delay in this reply - I've been hibernating for a bit.

np.

thanks,

C.

2006-11-22 21:32:07

by Andrew de Quincey

[permalink] [raw]
Subject: Re: [v4l-dvb-maintainer] Re: [Devel] Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

On Wednesday 22 November 2006 14:56, Cedric Le Goater wrote:
> Andrew de Quincey wrote:
> > Hi - the conversion looks good to me.. I can't really offer any more
> > constructive suggestions beyond what Cedric has already said.
>
> ok. so, should we just resend a refreshed version of the patch when 2.6.19
> comes out ?

Yeah - sounds good.

> > Theres another thread in dvb_ca_en50221.c that could be converted as well
> > though, hint hint ;)
>
> ok ok :) i'll look at it ...

heh :)

2006-12-12 22:59:38

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

Andrew de Quincey <[email protected]> writes:

> [snip]
>
>> correct, will fix that up in the next round
>>
>> thanks for the feedback,
>> Herbert
>
> Hi - the conversion looks good to me.. I can't really offer any more
> constructive suggestions beyond what Cedric has already said.
>
> Theres another thread in dvb_ca_en50221.c that could be converted as well
> though, hint hint ;)
>
> Apologies for the delay in this reply - I've been hibernating for a bit.

Guys where are we at on this conversion?

Eric

2006-12-12 23:14:49

by Herbert Poetzl

[permalink] [raw]
Subject: Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

On Tue, Dec 12, 2006 at 03:58:16PM -0700, Eric W. Biederman wrote:
> Andrew de Quincey <[email protected]> writes:
>
> > [snip]
> >
> >> correct, will fix that up in the next round
> >>
> >> thanks for the feedback,
> >> Herbert
> >
> > Hi - the conversion looks good to me.. I can't really offer any more
> > constructive suggestions beyond what Cedric has already said.
> >
> > Theres another thread in dvb_ca_en50221.c that could be converted as well
> > though, hint hint ;)
> >
> > Apologies for the delay in this reply - I've been hibernating for a bit.
>
> Guys where are we at on this conversion?

I can take a look at it in the next few days, but
I have no hardware to test that, so it would be
good to get in contact with somebody who does

best,
Herbert

> Eric

2006-12-13 15:55:21

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [Devel] Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

Herbert Poetzl wrote:
> On Tue, Dec 12, 2006 at 03:58:16PM -0700, Eric W. Biederman wrote:
>> Andrew de Quincey <[email protected]> writes:
>>
>>> [snip]
>>>
>>>> correct, will fix that up in the next round
>>>>
>>>> thanks for the feedback,
>>>> Herbert
>>> Hi - the conversion looks good to me.. I can't really offer any more
>>> constructive suggestions beyond what Cedric has already said.
>>>
>>> Theres another thread in dvb_ca_en50221.c that could be converted as well
>>> though, hint hint ;)
>>>
>>> Apologies for the delay in this reply - I've been hibernating for a bit.
>> Guys where are we at on this conversion?

I said I would but I didn't have much time yet?

however, I didn't see any big issue. no semaphore like in dvb_frontend.c.
just some kill(0) that looks harmless.

> I can take a look at it in the next few days, but
> I have no hardware to test that, so it would be
> good to get in contact with somebody who does

neither do I.

below is an updated version of herbert's dvb_frontend.c patch for 2.6.19-mm1

thanks,

C.

From: Herbert Poetzl <[email protected]>

Okay, as I promised, I had a first shot at the
dvb kernel_thread to kthread API port, and here
is the result, which is running fine here since
yesterday, including module load/unload and
software suspend (which doesn't work as expected
with or without this patch :), I didn't convert
the dvb_ca_en50221 as I do not have such an
interface, but if the conversion process is fine
with the v4l-dvb maintainers, it should not be
a problem to send a patch for that too ...

best,
Herbert

Signed-off-by: Herbert Poetzl <[email protected]>

---
drivers/media/dvb/dvb-core/dvb_frontend.c | 69 ++++++++++--------------------
drivers/media/dvb/ttpci/av7110.c | 29 +++++-------
drivers/media/dvb/ttpci/av7110.h | 1
3 files changed, 37 insertions(+), 62 deletions(-)

Index: 2.6.19-mm1/drivers/media/dvb/dvb-core/dvb_frontend.c
===================================================================
--- 2.6.19-mm1.orig/drivers/media/dvb/dvb-core/dvb_frontend.c
+++ 2.6.19-mm1/drivers/media/dvb/dvb-core/dvb_frontend.c
@@ -36,6 +36,7 @@
#include <linux/list.h>
#include <linux/freezer.h>
#include <linux/jiffies.h>
+#include <linux/kthread.h>
#include <asm/processor.h>

#include "dvb_frontend.h"
@@ -100,7 +101,7 @@ struct dvb_frontend_private {
struct semaphore sem;
struct list_head list_head;
wait_queue_head_t wait_queue;
- pid_t thread_pid;
+ struct task_struct *thread;
unsigned long release_jiffies;
unsigned int exit;
unsigned int wakeup;
@@ -508,19 +509,11 @@ static int dvb_frontend_thread(void *dat
struct dvb_frontend *fe = data;
struct dvb_frontend_private *fepriv = fe->frontend_priv;
unsigned long timeout;
- char name [15];
fe_status_t s;
struct dvb_frontend_parameters *params;

dprintk("%s\n", __FUNCTION__);

- snprintf (name, sizeof(name), "kdvb-fe-%i", fe->dvb->num);
-
- lock_kernel();
- daemonize(name);
- sigfillset(&current->blocked);
- unlock_kernel();
-
fepriv->check_wrapped = 0;
fepriv->quality = 0;
fepriv->delay = 3*HZ;
@@ -534,14 +527,16 @@ static int dvb_frontend_thread(void *dat
up(&fepriv->sem); /* is locked when we enter the thread... */

timeout = wait_event_interruptible_timeout(fepriv->wait_queue,
- dvb_frontend_should_wakeup(fe),
- fepriv->delay);
- if (0 != dvb_frontend_is_exiting(fe)) {
+ dvb_frontend_should_wakeup(fe) || kthread_should_stop(),
+ fepriv->delay);
+
+ if (kthread_should_stop() || dvb_frontend_is_exiting(fe)) {
/* got signal or quitting */
break;
}

- try_to_freeze();
+ if (try_to_freeze())
+ continue;

if (down_interruptible(&fepriv->sem))
break;
@@ -591,7 +586,7 @@ static int dvb_frontend_thread(void *dat
fe->ops.sleep(fe);
}

- fepriv->thread_pid = 0;
+ fepriv->thread = NULL;
mb();

dvb_frontend_wakeup(fe);
@@ -600,7 +595,6 @@ static int dvb_frontend_thread(void *dat

static void dvb_frontend_stop(struct dvb_frontend *fe)
{
- unsigned long ret;
struct dvb_frontend_private *fepriv = fe->frontend_priv;

dprintk ("%s\n", __FUNCTION__);
@@ -608,33 +602,17 @@ static void dvb_frontend_stop(struct dvb
fepriv->exit = 1;
mb();

- if (!fepriv->thread_pid)
+ if (!fepriv->thread)
return;

- /* check if the thread is really alive */
- if (kill_proc(fepriv->thread_pid, 0, 1) == -ESRCH) {
- printk("dvb_frontend_stop: thread PID %d already died\n",
- fepriv->thread_pid);
- /* make sure the mutex was not held by the thread */
- init_MUTEX (&fepriv->sem);
- return;
- }
-
- /* wake up the frontend thread, so it notices that fe->exit == 1 */
- dvb_frontend_wakeup(fe);
-
- /* wait until the frontend thread has exited */
- ret = wait_event_interruptible(fepriv->wait_queue,0 == fepriv->thread_pid);
- if (-ERESTARTSYS != ret) {
- fepriv->state = FESTATE_IDLE;
- return;
- }
+ kthread_stop(fepriv->thread);
+ init_MUTEX (&fepriv->sem);
fepriv->state = FESTATE_IDLE;

/* paranoia check in case a signal arrived */
- if (fepriv->thread_pid)
- printk("dvb_frontend_stop: warning: thread PID %d won't exit\n",
- fepriv->thread_pid);
+ if (fepriv->thread)
+ printk("dvb_frontend_stop: warning: thread %p won't exit\n",
+ fepriv->thread);
}

s32 timeval_usec_diff(struct timeval lasttime, struct timeval curtime)
@@ -684,10 +662,11 @@ static int dvb_frontend_start(struct dvb
{
int ret;
struct dvb_frontend_private *fepriv = fe->frontend_priv;
+ struct task_struct *fe_thread;

dprintk ("%s\n", __FUNCTION__);

- if (fepriv->thread_pid) {
+ if (fepriv->thread) {
if (!fepriv->exit)
return 0;
else
@@ -701,18 +680,18 @@ static int dvb_frontend_start(struct dvb

fepriv->state = FESTATE_IDLE;
fepriv->exit = 0;
- fepriv->thread_pid = 0;
+ fepriv->thread = NULL;
mb();

- ret = kernel_thread (dvb_frontend_thread, fe, 0);
-
- if (ret < 0) {
- printk("dvb_frontend_start: failed to start kernel_thread (%d)\n", ret);
+ fe_thread = kthread_run(dvb_frontend_thread, fe,
+ "kdvb-fe-%i", fe->dvb->num);
+ if (IS_ERR(fe_thread)) {
+ ret = PTR_ERR(fe_thread);
+ printk("dvb_frontend_start: failed to start kthread (%d)\n", ret);
up(&fepriv->sem);
return ret;
}
- fepriv->thread_pid = ret;
-
+ fepriv->thread = fe_thread;
return 0;
}

Index: 2.6.19-mm1/drivers/media/dvb/ttpci/av7110.c
===================================================================
--- 2.6.19-mm1.orig/drivers/media/dvb/ttpci/av7110.c
+++ 2.6.19-mm1/drivers/media/dvb/ttpci/av7110.c
@@ -51,6 +51,7 @@
#include <linux/firmware.h>
#include <linux/crc32.h>
#include <linux/i2c.h>
+#include <linux/kthread.h>

#include <asm/system.h>

@@ -223,11 +224,10 @@ static void recover_arm(struct av7110 *a

static void av7110_arm_sync(struct av7110 *av7110)
{
- av7110->arm_rmmod = 1;
- wake_up_interruptible(&av7110->arm_wait);
+ if (av7110->arm_thread)
+ kthread_stop(av7110->arm_thread);

- while (av7110->arm_thread)
- msleep(1);
+ av7110->arm_thread = NULL;
}

static int arm_thread(void *data)
@@ -238,17 +238,11 @@ static int arm_thread(void *data)

dprintk(4, "%p\n",av7110);

- lock_kernel();
- daemonize("arm_mon");
- sigfillset(&current->blocked);
- unlock_kernel();
-
- av7110->arm_thread = current;
-
for (;;) {
timeout = wait_event_interruptible_timeout(av7110->arm_wait,
- av7110->arm_rmmod, 5 * HZ);
- if (-ERESTARTSYS == timeout || av7110->arm_rmmod) {
+ kthread_should_stop(), 5 * HZ);
+
+ if (-ERESTARTSYS == timeout || kthread_should_stop()) {
/* got signal or told to quit*/
break;
}
@@ -276,7 +270,6 @@ static int arm_thread(void *data)
av7110->arm_errors = 0;
}

- av7110->arm_thread = NULL;
return 0;
}

@@ -2338,6 +2331,7 @@ static int __devinit av7110_attach(struc
const int length = TS_WIDTH * TS_HEIGHT;
struct pci_dev *pdev = dev->pci;
struct av7110 *av7110;
+ struct task_struct *thread;
int ret, count = 0;

dprintk(4, "dev: %p\n", dev);
@@ -2622,9 +2616,12 @@ static int __devinit av7110_attach(struc
printk ("dvb-ttpci: Warning, firmware version 0x%04x is too old. "
"System might be unstable!\n", FW_VERSION(av7110->arm_app));

- ret = kernel_thread(arm_thread, (void *) av7110, 0);
- if (ret < 0)
+ thread = kthread_run(arm_thread, (void *) av7110, "arm_mon");
+ if (IS_ERR(thread)) {
+ ret = PTR_ERR(thread);
goto err_stop_arm_9;
+ }
+ av7110->arm_thread = thread;

/* set initial volume in mixer struct */
av7110->mixer.volume_left = volume;
Index: 2.6.19-mm1/drivers/media/dvb/ttpci/av7110.h
===================================================================
--- 2.6.19-mm1.orig/drivers/media/dvb/ttpci/av7110.h
+++ 2.6.19-mm1/drivers/media/dvb/ttpci/av7110.h
@@ -205,7 +205,6 @@ struct av7110 {
struct task_struct *arm_thread;
wait_queue_head_t arm_wait;
u16 arm_loops;
- int arm_rmmod;

void *debi_virt;
dma_addr_t debi_bus;

2007-01-24 19:37:37

by Cédric Le Goater

[permalink] [raw]
Subject: Re: [PATCH/RFC] kthread API conversion for dvb_frontend and av7110

Cedric Le Goater wrote:
> Andrew de Quincey wrote:
>
>> Hi - the conversion looks good to me.. I can't really offer any more
>> constructive suggestions beyond what Cedric has already said.
>
> ok. so, should we just resend a refreshed version of the patch when 2.6.19
> comes out ?
>
>> Theres another thread in dvb_ca_en50221.c that could be converted as well
>> though, hint hint ;)
>
> ok ok :) i'll look at it ...

Here's a try. Compiles and boots but I have no hardware to test the
patch :(

could we replace wait_event_interruptible_timeout() with
wait_event_timeout() ? I don't see who would signal the thread.

thanks,

C.


Signed-off-by: Cedric Le Goater <[email protected]>
---
drivers/media/dvb/dvb-core/dvb_ca_en50221.c | 59 ++++++++--------------------
1 file changed, 18 insertions(+), 41 deletions(-)

Index: 2.6.20-rc4-mm1/drivers/media/dvb/dvb-core/dvb_ca_en50221.c
===================================================================
--- 2.6.20-rc4-mm1.orig/drivers/media/dvb/dvb-core/dvb_ca_en50221.c
+++ 2.6.20-rc4-mm1/drivers/media/dvb/dvb-core/dvb_ca_en50221.c
@@ -37,6 +37,7 @@
#include <linux/delay.h>
#include <linux/spinlock.h>
#include <linux/sched.h>
+#include <linux/kthread.h>

#include "dvb_ca_en50221.h"
#include "dvb_ringbuffer.h"
@@ -140,14 +141,11 @@ struct dvb_ca_private {
wait_queue_head_t wait_queue;

/* PID of the monitoring thread */
- pid_t thread_pid;
+ struct task_struct* thread;

/* Wait queue used when shutting thread down */
wait_queue_head_t thread_queue;

- /* Flag indicating when thread should exit */
- unsigned int exit:1;
-
/* Flag indicating if the CA device is open */
unsigned int open:1;

@@ -916,8 +914,6 @@ static int dvb_ca_en50221_thread_should_
ca->wakeup = 0;
return 1;
}
- if (ca->exit)
- return 1;

return 0;
}
@@ -982,7 +978,6 @@ static void dvb_ca_en50221_thread_update
static int dvb_ca_en50221_thread(void *data)
{
struct dvb_ca_private *ca = data;
- char name[15];
int slot;
int flags;
int status;
@@ -991,28 +986,19 @@ static int dvb_ca_en50221_thread(void *d

dprintk("%s\n", __FUNCTION__);

- /* setup kernel thread */
- snprintf(name, sizeof(name), "kdvb-ca-%i:%i", ca->dvbdev->adapter->num, ca->dvbdev->id);
-
- lock_kernel();
- daemonize(name);
- sigfillset(&current->blocked);
- unlock_kernel();
-
/* choose the correct initial delay */
dvb_ca_en50221_thread_update_delay(ca);

/* main loop */
- while (!ca->exit) {
+ while (1) {
/* sleep for a bit */
if (!ca->wakeup) {
- flags = wait_event_interruptible_timeout(ca->thread_queue,
- dvb_ca_en50221_thread_should_wakeup(ca),
- ca->delay);
- if ((flags == -ERESTARTSYS) || ca->exit) {
- /* got signal or quitting */
+ flags = wait_event_interruptible_timeout(
+ ca->thread_queue,
+ dvb_ca_en50221_thread_should_wakeup(ca) || kthread_should_stop(),
+ ca->delay);
+ if ((flags == -ERESTARTSYS) || kthread_should_stop())
break;
- }
}
ca->wakeup = 0;

@@ -1182,9 +1168,8 @@ static int dvb_ca_en50221_thread(void *d
}

/* completed */
- ca->thread_pid = 0;
+ ca->thread = NULL;
mb();
- wake_up_interruptible(&ca->thread_queue);
return 0;
}

@@ -1663,6 +1648,7 @@ int dvb_ca_en50221_init(struct dvb_adapt
int ret;
struct dvb_ca_private *ca = NULL;
int i;
+ struct task_struct *thread;

dprintk("%s\n", __FUNCTION__);

@@ -1682,9 +1668,8 @@ int dvb_ca_en50221_init(struct dvb_adapt
goto error;
}
init_waitqueue_head(&ca->wait_queue);
- ca->thread_pid = 0;
+ ca->thread = NULL;
init_waitqueue_head(&ca->thread_queue);
- ca->exit = 0;
ca->open = 0;
ca->wakeup = 0;
ca->next_read_slot = 0;
@@ -1711,13 +1696,14 @@ int dvb_ca_en50221_init(struct dvb_adapt

/* create a kthread for monitoring this CA device */

- ret = kernel_thread(dvb_ca_en50221_thread, ca, 0);
-
- if (ret < 0) {
+ thread = kthread_run(dvb_ca_en50221_thread, ca, "kdvb-ca-%i:%i",
+ ca->dvbdev->adapter->num, ca->dvbdev->id);
+ if (IS_ERR(thread)) {
+ ret = PTR_ERR(thread);
printk("dvb_ca_init: failed to start kernel_thread (%d)\n", ret);
goto error;
}
- ca->thread_pid = ret;
+ ca->thread = thread;
return 0;

error:
@@ -1748,17 +1734,8 @@ void dvb_ca_en50221_release(struct dvb_c
dprintk("%s\n", __FUNCTION__);

/* shutdown the thread if there was one */
- if (ca->thread_pid) {
- if (kill_proc(ca->thread_pid, 0, 1) == -ESRCH) {
- printk("dvb_ca_release adapter %d: thread PID %d already died\n",
- ca->dvbdev->adapter->num, ca->thread_pid);
- } else {
- ca->exit = 1;
- mb();
- dvb_ca_en50221_thread_wakeup(ca);
- wait_event_interruptible(ca->thread_queue, ca->thread_pid == 0);
- }
- }
+ if (ca->thread)
+ kthread_stop(ca->thread);

for (i = 0; i < ca->slot_count; i++) {
dvb_ca_en50221_slot_shutdown(ca, i);