NetBSD Problem Report #36673
From yamt@mwd.biglobe.ne.jp Sat Jul 21 11:31:39 2007
Return-Path: <yamt@mwd.biglobe.ne.jp>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
by narn.NetBSD.org (Postfix) with ESMTP id 04D8D63B93E
for <gnats-bugs@gnats.NetBSD.org>; Sat, 21 Jul 2007 11:31:39 +0000 (UTC)
Message-Id: <20070721113137.1113711702@yamt.dyndns.org>
Date: Sat, 21 Jul 2007 20:31:37 +0900 (JST)
From: yamt@mwd.biglobe.ne.jp
Reply-To: yamt@mwd.biglobe.ne.jp
To: gnats-bugs@NetBSD.org
Subject: dubious code in sysmon_envsys
X-Send-Pr-Version: 3.95
>Number: 36673
>Category: kern
>Synopsis: dubious code is sysmon_envsys
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: kern-bug-people
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Sat Jul 21 11:35:00 +0000 2007
>Closed-Date: Sat Nov 14 23:38:56 +0000 2009
>Last-Modified: Sat Nov 14 23:38:56 +0000 2009
>Originator: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
>Release: NetBSD 4.99.24
>Organization:
>Environment:
Architecture: i386
Machine: i386
>Description:
- sme_register_sensorname doesn't need to handle EMPTY case specially.
- sme_make_dictionary should fully set up a dictionary before
adding it to the array. (instead of leaving a partial dictionary
on errors.)
same for sysmon_envsys_createplist and its array.
- ENVSYS_FDUPDESC seems broken for compat ioctls.
- what global mutexes like sme_mtx protect is not clear.
- 'obj' argument of sme_sensor_upstring (and similar functions)
is useless.
- '--sme->sme_nsensors' in sme_update_dictionary seems quite dubious.
- sysmon_envsys_createplist seems to ignore most errors
unless it was for the last sensor in sme_sensor_data[].
i don't understand why the last one is special.
>How-To-Repeat:
code inspection.
>Fix:
>Release-Note:
>Audit-Trail:
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sat, 21 Jul 2007 14:16:45 +0200
On Sat, 21 Jul 2007 11:35:00 +0000 (UTC)
yamt@mwd.biglobe.ne.jp wrote:
> - sme_register_sensorname doesn't need to handle EMPTY case
> specially.
Fixed.
> - sme_make_dictionary should fully set up a dictionary before
> adding it to the array. (instead of leaving a partial
> dictionary on errors.)
> same for sysmon_envsys_createplist and its array.
Are you saying that only one error should destroy the whole array?
and why?
> - ENVSYS_FDUPDESC seems broken for compat ioctls.
It seems so and I'll fix.
> - what global mutexes like sme_mtx protect is not clear.
Not clear why? can you explain more?
> - 'obj' argument of sme_sensor_upstring (and similar
> functions) is useless.
Fixed.
> - '--sme->sme_nsensors' in sme_update_dictionary seems quite
> dubious.
Dubious, why? how do you want to skip sensors with duplicate
description, otherwise? let me know if you know any other way and I'll
fix.
> - sysmon_envsys_createplist seems to ignore most errors
> unless it was for the last sensor in sme_sensor_data[].
> i don't understand why the last one is special.
I don't understand what you are reporting here.
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sat, 21 Jul 2007 17:33:02 +0200
On Sat, 21 Jul 2007 11:35:00 +0000 (UTC)
yamt@mwd.biglobe.ne.jp wrote:
Ok, I've made a few changes... let's see what's missing now.
> - sme_register_sensorname doesn't need to handle EMPTY case
> specially.
Fixed.
> - sme_make_dictionary should fully set up a dictionary before
> adding it to the array. (instead of leaving a partial
> dictionary on errors.)
> same for sysmon_envsys_createplist and its array.
>
> - ENVSYS_FDUPDESC seems broken for compat ioctls.
Fixed.
> - what global mutexes like sme_mtx protect is not clear.
I believe they are used ok, correct me if I'm wrong.
> - 'obj' argument of sme_sensor_upstring (and similar
> functions) is useless.
Fixed.
> - '--sme->sme_nsensors' in sme_update_dictionary seems quite
> dubious.
This is the only one that I'm not sure why you are complaining, it
works ok... do you know any other way to skip sensors with duplicate
descriptions?
> - sysmon_envsys_createplist seems to ignore most errors
> unless it was for the last sensor in sme_sensor_data[].
> i don't understand why the last one is special.
It returns the error if there's any now.
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sun, 22 Jul 2007 00:45:02 +0900 (JST)
> > - sme_make_dictionary should fully set up a dictionary before
> > adding it to the array. (instead of leaving a partial
> > dictionary on errors.)
> > same for sysmon_envsys_createplist and its array.
>
> Are you saying that only one error should destroy the whole array?
> and why?
because it's the simplest way to handle errors.
and it's clearer, at least than leaving partially initialized data.
> > - what global mutexes like sme_mtx protect is not clear.
>
> Not clear why? can you explain more?
because it isn't documented. it might be clear to you, but not to me.
> > - '--sme->sme_nsensors' in sme_update_dictionary seems quite
> > dubious.
>
> Dubious, why?
because i don't understand why you want to modify sme_nsensors here.
besides, the current code:
for (i = 0; i < sme->sme_nsensors; i++) {
edata = &sme->sme_sensor_data[i];
if (edata->flags & ENVSYS_FDUPDESC) {
--sme->sme_nsensors;
--i;
continue;
}
"--i; continue;" means "i" will be the same value in the next iteration of
the loop. so it will look at the same "edata", and the ENVSYS_FDUPDESC
condition will always to be true until the loop exits.
so it's effectively same as:
for (i = 0; i < sme->sme_nsensors; i++) {
edata = &sme->sme_sensor_data[i];
if (edata->flags & ENVSYS_FDUPDESC) {
sme->sme_nsensors = i;
break;
}
is it what you intend?
> how do you want to skip sensors with duplicate
> description, otherwise? let me know if you know any other way and I'll
> fix.
define "skip sensors" precisely. otherwise i can't say how you can do it.
i often feel you are assuming your intention is obvious to your peer.
actually it isn't, especially when the code seems broken.
> > - sysmon_envsys_createplist seems to ignore most errors
> > unless it was for the last sensor in sme_sensor_data[].
> > i don't understand why the last one is special.
>
> I don't understand what you are reporting here.
suppose that sme->sme_nsensors == 2.
if an error happens for sme->sme_sensor_data[0], it will not be
returned to the caller of sysmon_envsys_createplist. (unless it was EINVAL)
otoh, if an error happens for sme->sme_sensor_data[1], it will be
returned to the caller of sysmon_envsys_createplist.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sat, 21 Jul 2007 18:03:43 +0200
On Sun, 22 Jul 2007 00:45:02 +0900 (JST)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> because it's the simplest way to handle errors.
> and it's clearer, at least than leaving partially initialized data.
Ok, I think you are right. I think we should have at least three or
four required objects in a dictionary to be created:
- state
- type
- description
- something else?
> because it isn't documented. it might be clear to you, but not to me.
Do I have to explain how the locking works in the code? I think that if
you see the code, you can see how it works.
There is source less documented than mine and I don't complain because
there's not stuff documented...
> because i don't understand why you want to modify sme_nsensors here.
>
> besides, the current code:
>
> for (i = 0; i < sme->sme_nsensors; i++) {
> edata = &sme->sme_sensor_data[i];
> if (edata->flags & ENVSYS_FDUPDESC) {
> --sme->sme_nsensors;
> --i;
> continue;
> }
>
> "--i; continue;" means "i" will be the same value in the next
> iteration of the loop. so it will look at the same "edata", and the
> ENVSYS_FDUPDESC condition will always to be true until the loop exits.
> so it's effectively same as:
>
> for (i = 0; i < sme->sme_nsensors; i++) {
> edata = &sme->sme_sensor_data[i];
> if (edata->flags & ENVSYS_FDUPDESC) {
> sme->sme_nsensors = i;
> break;
> }
>
> is it what you intend?
But that would mean that if you break and it's not the last sensor,
next sensors will be skipped... right? this is not what we want to make
it show all valid sensors.
> suppose that sme->sme_nsensors == 2.
> if an error happens for sme->sme_sensor_data[0], it will not be
> returned to the caller of sysmon_envsys_createplist. (unless it was
> EINVAL) otoh, if an error happens for sme->sme_sensor_data[1], it
> will be returned to the caller of sysmon_envsys_createplist.
You are right. I changed the code to check for all errors, let me know
if there are more problems.
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sat, 21 Jul 2007 18:19:35 +0200
On Sat, 21 Jul 2007 16:05:03 +0000 (UTC)
Juan RP <juan@xtrarom.org> wrote:
> > because i don't understand why you want to modify sme_nsensors
> > here.
> >
> > besides, the current code:
> >
> > for (i = 0; i < sme->sme_nsensors; i++) {
> > edata = &sme->sme_sensor_data[i];
> > if (edata->flags & ENVSYS_FDUPDESC) {
> > --sme->sme_nsensors;
> > --i;
> > continue;
> > }
> >
> > "--i; continue;" means "i" will be the same value in the next
> > iteration of the loop. so it will look at the same "edata", and
> > the ENVSYS_FDUPDESC condition will always to be true until the
> > loop exits. so it's effectively same as:
> >
> > for (i = 0; i < sme->sme_nsensors; i++) {
> > edata = &sme->sme_sensor_data[i];
> > if (edata->flags & ENVSYS_FDUPDESC) {
> > sme->sme_nsensors = i;
> > break;
> > }
> >
> > is it what you intend?
>
> But that would mean that if you break and it's not the last sensor,
> next sensors will be skipped... right? this is not what we want to
> make it show all valid sensors.
I removed this bogus '--i' and all works as expected. I think this was
required in a previous implementation that I had, but not needed since
I added a SLIST to handle the sensor descriptions.
> > suppose that sme->sme_nsensors == 2.
> > if an error happens for sme->sme_sensor_data[0], it will not be
> > returned to the caller of sysmon_envsys_createplist. (unless it
> > was EINVAL) otoh, if an error happens for sme->sme_sensor_data[1],
> > it will be returned to the caller of sysmon_envsys_createplist.
>
> You are right. I changed the code to check for all errors, let me
> know if there are more problems.
Thanks. I think there's only one point to fix your PR right now?
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sun, 22 Jul 2007 01:26:03 +0900 (JST)
> > because it isn't documented. it might be clear to you, but not to me.
>
> Do I have to explain how the locking works in the code? I think that if
> you see the code, you can see how it works.
>
> There is source less documented than mine and I don't complain because
> there's not stuff documented...
if you want someone understand (and possibly fix) your code,
it's better to explain unless it's obvious.
to me, the locking here seems far from obvious. for example, to me,
- it isn't clear which of sme_mtx or sme_list_mtx protects sysmon_envsys_list.
- it isn't clear if it's safe to drop sme_mtx in the middle of
sme_make_dictionary.
> > because i don't understand why you want to modify sme_nsensors here.
> >
> > besides, the current code:
> >
> > for (i = 0; i < sme->sme_nsensors; i++) {
> > edata = &sme->sme_sensor_data[i];
> > if (edata->flags & ENVSYS_FDUPDESC) {
> > --sme->sme_nsensors;
> > --i;
> > continue;
> > }
> >
> > "--i; continue;" means "i" will be the same value in the next
> > iteration of the loop. so it will look at the same "edata", and the
> > ENVSYS_FDUPDESC condition will always to be true until the loop exits.
> > so it's effectively same as:
> >
> > for (i = 0; i < sme->sme_nsensors; i++) {
> > edata = &sme->sme_sensor_data[i];
> > if (edata->flags & ENVSYS_FDUPDESC) {
> > sme->sme_nsensors = i;
> > break;
> > }
> >
> > is it what you intend?
>
> But that would mean that if you break and it's not the last sensor,
> next sensors will be skipped... right? this is not what we want to make
> it show all valid sensors.
although i'm not sure what you want to do here,
a usual way to "skip" an entry in a loop is just "continue".
for (i = 0; i < sme->sme_nsensors; i++) {
edata = &sme->sme_sensor_data[i];
if (edata->flags & ENVSYS_FDUPDESC) {
continue;
}
can you explain why you want to modify sme_nsensors here?
it isn't obvious to me.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sat, 21 Jul 2007 18:59:38 +0200
On Sat, 21 Jul 2007 16:30:02 +0000 (UTC)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> to me, the locking here seems far from obvious. for example, to me,
>
> - it isn't clear which of sme_mtx or sme_list_mtx protects
> sysmon_envsys_list.
This should be sme_list_mtx for the LIST. I fixed most cases, but
perhaps there are still some, I'll fix.
> - it isn't clear if it's safe to drop sme_mtx in the middle of
> sme_make_dictionary.
Can I allocate/deallocate memory with kmem(9) with a mutex held?
someone said that it's not possible because it could cause a deadlock
or something.
> although i'm not sure what you want to do here,
> a usual way to "skip" an entry in a loop is just "continue".
>
> for (i = 0; i < sme->sme_nsensors; i++) {
> edata = &sme->sme_sensor_data[i];
> if (edata->flags & ENVSYS_FDUPDESC) {
> continue;
> }
>
> can you explain why you want to modify sme_nsensors here?
> it isn't obvious to me.
Sure. In sme_make_dictionary the sensors with a duplicate description
do not create another dictionary in the array, so the number of
dictionaries in the array doesn't match with sme_nsensors.
So let's see this code in sme_update_dictionary():
for (i = 0; i < sme->sme_nsensors; i++) {
edata = &sme->sme_sensor_data[i];
if (edata->flags & ENVSYS_FDUPDESC) {
--sme->sme_nsensors;
continue;
}
With the code in sys/lkm/misc/envsys2, there are 15 sensors in total
and sensors with index 13 and 14 have a duplicate description.
If sme_nsensors is not decreased the following code:
/* retrieve sensor's dictionary. */
dict = prop_array_get(array, i);
if (prop_object_type(dict) != PROP_TYPE_DICTIONARY) {
DPRINTF(("%s: not a dictionary (%d:%s)\n",
__func__, edata->sensor, sme->sme_name));
return EINVAL;
}
will fail, because 'i' will still try to access to the index specified
in sme_nsensors, but the dictionary doesn't exist and it will fail.
Is it clear for you now?
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sun, 22 Jul 2007 02:33:34 +0900 (JST)
> On Sat, 21 Jul 2007 16:30:02 +0000 (UTC)
> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
>
> > to me, the locking here seems far from obvious. for example, to me,
> >
> > - it isn't clear which of sme_mtx or sme_list_mtx protects
> > sysmon_envsys_list.
>
> This should be sme_list_mtx for the LIST. I fixed most cases, but
> perhaps there are still some, I'll fix.
then, sysmonioctl_envsys seems broken.
> > - it isn't clear if it's safe to drop sme_mtx in the middle of
> > sme_make_dictionary.
>
> Can I allocate/deallocate memory with kmem(9) with a mutex held?
> someone said that it's not possible because it could cause a deadlock
> or something.
it depends.
what i want to know is if it's safe to drop the lock here,
rather than a reason why you want to drop it...
> > although i'm not sure what you want to do here,
> > a usual way to "skip" an entry in a loop is just "continue".
> >
> > for (i = 0; i < sme->sme_nsensors; i++) {
> > edata = &sme->sme_sensor_data[i];
> > if (edata->flags & ENVSYS_FDUPDESC) {
> > continue;
> > }
> >
> > can you explain why you want to modify sme_nsensors here?
> > it isn't obvious to me.
>
> Sure. In sme_make_dictionary the sensors with a duplicate description
> do not create another dictionary in the array, so the number of
> dictionaries in the array doesn't match with sme_nsensors.
>
> So let's see this code in sme_update_dictionary():
>
> for (i = 0; i < sme->sme_nsensors; i++) {
> edata = &sme->sme_sensor_data[i];
> if (edata->flags & ENVSYS_FDUPDESC) {
> --sme->sme_nsensors;
> continue;
> }
>
> With the code in sys/lkm/misc/envsys2, there are 15 sensors in total
> and sensors with index 13 and 14 have a duplicate description.
>
> If sme_nsensors is not decreased the following code:
>
> /* retrieve sensor's dictionary. */
> dict = prop_array_get(array, i);
> if (prop_object_type(dict) != PROP_TYPE_DICTIONARY) {
> DPRINTF(("%s: not a dictionary (%d:%s)\n",
> __func__, edata->sensor, sme->sme_name));
> return EINVAL;
> }
>
> will fail, because 'i' will still try to access to the index specified
> in sme_nsensors, but the dictionary doesn't exist and it will fail.
>
> Is it clear for you now?
what happens when you call sme_update_dictionary again?
you decrease sme_nsensors again erroneously, don't you?
i guess it's clearer to always use same indexes for sme_sensor_data and
prop_array. ie. use prop_array_set instead of prop_array_add.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sat, 21 Jul 2007 19:39:23 +0200
On Sat, 21 Jul 2007 17:35:01 +0000 (UTC)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> then, sysmonioctl_envsys seems broken.
I fixed it, please check latest code.
> it depends.
>
> what i want to know is if it's safe to drop the lock here,
> rather than a reason why you want to drop it...
Do you think is it safe? well, I don't know. Do you have any idea
how to change it to be safe?
> what happens when you call sme_update_dictionary again?
> you decrease sme_nsensors again erroneously, don't you?
>
> i guess it's clearer to always use same indexes for sme_sensor_data
> and prop_array. ie. use prop_array_set instead of prop_array_add.
Oh right... I didn't think about this case. I'll fix now, thank you.
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673: dubious code in sysmon_envsys
Date: Sun, 22 Jul 2007 02:53:02 +0900 (JST)
> > what i want to know is if it's safe to drop the lock here,
> > rather than a reason why you want to drop it...
>
> Do you think is it safe? well, I don't know.
i don't know.
> Do you have any idea
> how to change it to be safe?
if i were you, i'd do it like the following.
/* setup new data without any lock held
(create and fill a dictionary, etc) */
mutex_enter(global_mutex);
/* attach the above data into the global structure,
so that it's visible from other threads.
(sme_register_sensorname, prop_array_add) */
mutex_exit(global_mutex);
YAMAMOTO Takashi
Responsible-Changed-From-To: kern-bug-people->xtraeme
Responsible-Changed-By: xtraeme@netbsd.org
Responsible-Changed-When: Sun, 22 Jul 2007 11:32:11 +0000
Responsible-Changed-Why:
mine
State-Changed-From-To: open->feedback
State-Changed-By: xtraeme@netbsd.org
State-Changed-When: Sun, 22 Jul 2007 11:32:11 +0000
State-Changed-Why:
I believe most things you reported were fixed, can you please check
again? thanks.
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: gnats-bugs@NetBSD.org
Cc: xtraeme@NetBSD.org, kern-bug-people@netbsd.org,
netbsd-bugs@netbsd.org, gnats-admin@netbsd.org, xtraeme@netbsd.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Sun, 22 Jul 2007 20:54:47 +0900 (JST)
> State-Changed-From-To: open->feedback
> State-Changed-By: xtraeme@netbsd.org
> State-Changed-When: Sun, 22 Jul 2007 11:32:11 +0000
> State-Changed-Why:
> I believe most things you reported were fixed, can you please check
> again? thanks.
please check it by yourself.
at least, this one doesn't seems to be fixed.
> - sme_make_dictionary should fully set up a dictionary before
> adding it to the array. (instead of leaving a partial dictionary
> on errors.)
> same for sysmon_envsys_createplist and its array.
YAMAMOTO Takashi
State-Changed-From-To: feedback->closed
State-Changed-By: xtraeme@netbsd.org
State-Changed-When: Sun, 22 Jul 2007 18:18:27 +0000
State-Changed-Why:
I just fixed your last point, thanks for your PR.
From: Juan Romero Pardines <xtraeme@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc:
Subject: PR/36673 CVS commit: src/sys
Date: Sun, 22 Jul 2007 18:17:03 +0000 (UTC)
Module Name: src
Committed By: xtraeme
Date: Sun Jul 22 18:17:03 UTC 2007
Modified Files:
src/sys/dev/sysmon: sysmon_envsys.c sysmon_envsysvar.h
src/sys/sys: envsys.h
Log Message:
Fix the last point reported by yamt in PR kern/36673:
- There must be three mandatory objects in a sensor to be valid and be
enabled on its dictionary: units, state and desc. If any of these
objects fails for some reason, its dictionary won't created in the
array and sensor will be marked with ENVSYS_FNOTVALID in edata->flags.
That means there won't never be a partial dictionary.
- Mark all invalid sensors and sensors with duplicate description with
ENVSYS_FNOTVALID and remove ENVSYS_FDUPDESC. The former is the generic
way to say that a sensor does not obey the rules and must be not
created.
- sysmon_envsys_createplist: if the array does not contain any object,
release the array and return the error.
And with all that stuff I can go to holidays and be happy.
To generate a diff of this commit:
cvs rdiff -r1.40 -r1.41 src/sys/dev/sysmon/sysmon_envsys.c
cvs rdiff -r1.10 -r1.11 src/sys/dev/sysmon/sysmon_envsysvar.h
cvs rdiff -r1.13 -r1.14 src/sys/sys/envsys.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: gnats-bugs@NetBSD.org
Cc: xtraeme@NetBSD.org, netbsd-bugs@netbsd.org,
gnats-admin@netbsd.org, xtraeme@netbsd.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Mon, 23 Jul 2007 10:59:46 +0900 (JST)
> Synopsis: dubious code is sysmon_envsys
>
> State-Changed-From-To: feedback->closed
> State-Changed-By: xtraeme@netbsd.org
> State-Changed-When: Sun, 22 Jul 2007 18:18:27 +0000
> State-Changed-Why:
> I just fixed your last point, thanks for your PR.
please re-open.
can you re-read the PR and code by yourself before closing it? thanks.
> - what global mutexes like sme_mtx protect is not clear.
> - sysmon_envsys_createplist seems to ignore most errors
> unless it was for the last sensor in sme_sensor_data[].
> i don't understand why the last one is special.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Mon, 23 Jul 2007 10:23:30 +0200
On Mon, 23 Jul 2007 10:59:46 +0900 (JST)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> please re-open.
> can you re-read the PR and code by yourself before closing it?
> thanks.
>
> > - what global mutexes like sme_mtx protect is not clear.
What's the problem with this?
> > - sysmon_envsys_createplist seems to ignore most errors
> > unless it was for the last sensor in sme_sensor_data[].
> > i don't understand why the last one is special.
Have you seen the changes I made yesterday? how is the current
code looking at the last error of sme_sensor_data now?
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
State-Changed-From-To: closed->open
State-Changed-By: xtraeme@netbsd.org
State-Changed-When: Mon, 23 Jul 2007 08:49:46 +0000
State-Changed-Why:
There's still one point that needs to be fixed:
- what global mutexes like sme_mtx protect is not clear.
From: Juan Romero Pardines <xtraeme@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc:
Subject: PR/36673 CVS commit: src/sys/dev/sysmon
Date: Mon, 23 Jul 2007 08:45:51 +0000 (UTC)
Module Name: src
Committed By: xtraeme
Date: Mon Jul 23 08:45:51 UTC 2007
Modified Files:
src/sys/dev/sysmon: sysmon_envsys.c
Log Message:
- sysmon_envsys_createplist: do not return the error of sme->sme_gtredata
in all cases to the caller, just continue and show a debug printf if it
fails, because it's not a critical error.
- Improve some debugging printfs.
Fixes another part of yamt@'s PR kern/36673, that I forgot to commit
yesterday.
To generate a diff of this commit:
cvs rdiff -r1.41 -r1.42 src/sys/dev/sysmon/sysmon_envsys.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Tue, 24 Jul 2007 07:55:24 +0900 (JST)
> On Mon, 23 Jul 2007 10:59:46 +0900 (JST)
> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
>
> > please re-open.
> > can you re-read the PR and code by yourself before closing it?
> > thanks.
> >
> > > - what global mutexes like sme_mtx protect is not clear.
>
> What's the problem with this?
are you asking how it isn't clear?
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Tue, 24 Jul 2007 00:58:17 +0200
On Tue, 24 Jul 2007 07:55:24 +0900 (JST)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> are you asking how it isn't clear?
I wanted to know what parts are wrong or how they can be improved.
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Tue, 24 Jul 2007 08:16:33 +0900 (JST)
> On Tue, 24 Jul 2007 07:55:24 +0900 (JST)
> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
>
> > are you asking how it isn't clear?
>
> I wanted to know what parts are wrong or how they can be improved.
the usages of these mutexes are not obvious or documented.
for example,
- i can't understand why sysmon_envsys_createplist acquires
sme_mtx in the loop.
- sysmon_envsys_register seems to assume sysmon_envsys_list is stable
while unlocking sme_list_mtx.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Tue, 24 Jul 2007 01:35:30 +0200
On Mon, 23 Jul 2007 23:20:03 +0000 (UTC)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> the usages of these mutexes are not obvious or documented.
>
> for example,
> - i can't understand why sysmon_envsys_createplist acquires
> sme_mtx in the loop.
I wanted to protect the sme_gtredata function callback, this one
is executed in the driver to refresh data in the sensors.
> - sysmon_envsys_register seems to assume sysmon_envsys_list is stable
> while unlocking sme_list_mtx.
Do you think the following patch is ok for you?
Index: sysmon_envsys.c
===================================================================
RCS file: /cvsroot/src/sys/dev/sysmon/sysmon_envsys.c,v
retrieving revision 1.43
diff -b -u -r1.43 sysmon_envsys.c
--- sysmon_envsys.c 23 Jul 2007 17:51:16 -0000 1.43
+++ sysmon_envsys.c 23 Jul 2007 23:30:18 -0000
@@ -483,7 +483,6 @@
mutex_enter(&sme_list_mtx);
LIST_FOREACH(lsme, &sysmon_envsys_list, sme_list) {
if (strcmp(lsme->sme_name, sme->sme_name) == 0) {
- mutex_exit(&sme_list_mtx);
error = EEXIST;
goto out;
}
@@ -496,16 +495,14 @@
sme->sme_fsensor = sysmon_envsys_next_sensor_index;
sysmon_envsys_next_sensor_index += sme->sme_nsensors;
#endif
- mutex_exit(&sme_list_mtx);
error = sysmon_envsys_createplist(sme);
if (!error) {
- mutex_enter(&sme_list_mtx);
LIST_INSERT_HEAD(&sysmon_envsys_list, sme, sme_list);
sme_uniqsensors = 0;
- mutex_exit(&sme_list_mtx);
}
out:
+ mutex_exit(&sme_mtx);
return error;
}
@@ -626,6 +623,9 @@
nsens = error = 0;
+ KASSERT(mutex_owned(&sme_list_mtx));
+ mutex_exit(&sme_list_mtx);
+
/* create the sysmon envsys device array. */
array = prop_array_create();
if (array == NULL)
One thing that is not clear to me is if I can allocate/deallocate memory
with a mutex held, can you please answer?
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Tue, 24 Jul 2007 01:53:25 +0200
On Mon, 23 Jul 2007 23:40:03 +0000 (UTC)
Juan RP <juan@xtrarom.org> wrote:
> Do you think the following patch is ok for you?
Erm... please ignore previous patch, what about this one?
Index: sysmon_envsys.c
===================================================================
RCS file: /cvsroot/src/sys/dev/sysmon/sysmon_envsys.c,v
retrieving revision 1.43
diff -b -u -r1.43 sysmon_envsys.c
--- sysmon_envsys.c 23 Jul 2007 17:51:16 -0000 1.43
+++ sysmon_envsys.c 23 Jul 2007 23:51:47 -0000
@@ -483,7 +483,6 @@
mutex_enter(&sme_list_mtx);
LIST_FOREACH(lsme, &sysmon_envsys_list, sme_list) {
if (strcmp(lsme->sme_name, sme->sme_name) == 0) {
- mutex_exit(&sme_list_mtx);
error = EEXIST;
goto out;
}
@@ -496,16 +495,14 @@
sme->sme_fsensor = sysmon_envsys_next_sensor_index;
sysmon_envsys_next_sensor_index += sme->sme_nsensors;
#endif
- mutex_exit(&sme_list_mtx);
error = sysmon_envsys_createplist(sme);
if (!error) {
- mutex_enter(&sme_list_mtx);
LIST_INSERT_HEAD(&sysmon_envsys_list, sme, sme_list);
sme_uniqsensors = 0;
- mutex_exit(&sme_list_mtx);
}
out:
+ mutex_exit(&sme_list_mtx);
return error;
}
@@ -626,6 +623,9 @@
nsens = error = 0;
+ KASSERT(mutex_owned(&sme_list_mtx));
+ mutex_exit(&sme_list_mtx);
+
/* create the sysmon envsys device array. */
array = prop_array_create();
if (array == NULL)
@@ -688,6 +688,7 @@
prop_object_release(array);
+ mutex_enter(&sme_list_mtx);
return error;
}
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 09:15:38 +0900 (JST)
> On Mon, 23 Jul 2007 23:20:03 +0000 (UTC)
> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
>
> > the usages of these mutexes are not obvious or documented.
> >
> > for example,
> > - i can't understand why sysmon_envsys_createplist acquires
> > sme_mtx in the loop.
>
> I wanted to protect the sme_gtredata function callback, this one
> is executed in the driver to refresh data in the sensors.
protect from what?
> One thing that is not clear to me is if I can allocate/deallocate memory
> with a mutex held, can you please answer?
as i already answered, it depends.
don't do it if you are not sure.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 12:58:12 +0200
On Wed, 25 Jul 2007 09:15:38 +0900 (JST)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> > On Mon, 23 Jul 2007 23:20:03 +0000 (UTC)
> > yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> >
> > > the usages of these mutexes are not obvious or documented.
> > >
> > > for example,
> > > - i can't understand why sysmon_envsys_createplist acquires
> > > sme_mtx in the loop.
> >
> > I wanted to protect the sme_gtredata function callback, this one
> > is executed in the driver to refresh data in the sensors.
>
> protect from what?
Protect from retrieve sensor data from multiple threads.
> > One thing that is not clear to me is if I can allocate/deallocate
> > memory with a mutex held, can you please answer?
>
> as i already answered, it depends.
> don't do it if you are not sure.
I'm not doing it on my code, but I would like to have more details than
"it depends". Depends of what? depends if is it a spin mutex or an
adaptive mutex?
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 13:02:48 +0200
On Mon, 23 Jul 2007 23:55:02 +0000 (UTC)
Juan RP <juan@xtrarom.org> wrote:
> Erm... please ignore previous patch, what about this one?
Can you comment about the last patch I posted?
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 20:44:33 +0900 (JST)
> On Mon, 23 Jul 2007 23:55:02 +0000 (UTC)
> Juan RP <juan@xtrarom.org> wrote:
>
> > Erm... please ignore previous patch, what about this one?
>
> Can you comment about the last patch I posted?
can you provide a diff with -p?
YAMAMOTO Takashi
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: gnats-bugs@NetBSD.org
Cc: xtraeme@NetBSD.org, gnats-admin@netbsd.org,
netbsd-bugs@netbsd.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 20:44:52 +0900 (JST)
> The following reply was made to PR kern/36673; it has been noted by GNATS.
>
> From: Juan RP <juan@xtrarom.org>
> To: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
> Cc: gnats-bugs@NetBSD.org
> Subject: Re: kern/36673 (dubious code is sysmon_envsys)
> Date: Wed, 25 Jul 2007 12:58:12 +0200
>
> On Wed, 25 Jul 2007 09:15:38 +0900 (JST)
> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
>
> > > On Mon, 23 Jul 2007 23:20:03 +0000 (UTC)
> > > yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> > >
> > > > the usages of these mutexes are not obvious or documented.
> > > >
> > > > for example,
> > > > - i can't understand why sysmon_envsys_createplist acquires
> > > > sme_mtx in the loop.
> > >
> > > I wanted to protect the sme_gtredata function callback, this one
> > > is executed in the driver to refresh data in the sensors.
> >
> > protect from what?
>
> Protect from retrieve sensor data from multiple threads.
is it visible from other threads at this point?
> > > One thing that is not clear to me is if I can allocate/deallocate
> > > memory with a mutex held, can you please answer?
> >
> > as i already answered, it depends.
> > don't do it if you are not sure.
>
> I'm not doing it on my code, but I would like to have more details than
> "it depends". Depends of what? depends if is it a spin mutex or an
> adaptive mutex?
it's unsafe if it can interfere pagedaemon, for example.
YAMAMOTO Takashi
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 14:35:35 +0200
On Wed, 25 Jul 2007 11:45:05 +0000 (UTC)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> is it visible from other threads at this point?
I think it's not visible.
> it's unsafe if it can interfere pagedaemon, for example.
Ok.
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
From: Juan RP <juan@xtrarom.org>
To: gnats-bugs@NetBSD.org
Cc: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Wed, 25 Jul 2007 14:36:48 +0200
On Wed, 25 Jul 2007 11:45:04 +0000 (UTC)
yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> can you provide a diff with -p?
Sure, here it is: (I didn't know about -p, it's very useful :-)
Index: sysmon_envsys.c
===================================================================
RCS file: /cvsroot/src/sys/dev/sysmon/sysmon_envsys.c,v
retrieving revision 1.43
diff -b -u -p -r1.43 sysmon_envsys.c
--- sysmon_envsys.c 23 Jul 2007 17:51:16 -0000 1.43
+++ sysmon_envsys.c 25 Jul 2007 12:34:15 -0000
@@ -483,7 +483,6 @@ sysmon_envsys_register(struct sysmon_env
mutex_enter(&sme_list_mtx);
LIST_FOREACH(lsme, &sysmon_envsys_list, sme_list) {
if (strcmp(lsme->sme_name, sme->sme_name) == 0) {
- mutex_exit(&sme_list_mtx);
error = EEXIST;
goto out;
}
@@ -496,16 +495,14 @@ sysmon_envsys_register(struct sysmon_env
sme->sme_fsensor = sysmon_envsys_next_sensor_index;
sysmon_envsys_next_sensor_index += sme->sme_nsensors;
#endif
- mutex_exit(&sme_list_mtx);
error = sysmon_envsys_createplist(sme);
if (!error) {
- mutex_enter(&sme_list_mtx);
LIST_INSERT_HEAD(&sysmon_envsys_list, sme, sme_list);
sme_uniqsensors = 0;
- mutex_exit(&sme_list_mtx);
}
out:
+ mutex_exit(&sme_list_mtx);
return error;
}
@@ -626,6 +623,9 @@ sysmon_envsys_createplist(struct sysmon_
nsens = error = 0;
+ KASSERT(mutex_owned(&sme_list_mtx));
+ mutex_exit(&sme_list_mtx);
+
/* create the sysmon envsys device array. */
array = prop_array_create();
if (array == NULL)
@@ -688,6 +688,7 @@ sysmon_envsys_createplist(struct sysmon_
prop_object_release(array);
+ mutex_enter(&sme_list_mtx);
return error;
}
--
Juan Romero Pardines - The NetBSD Project
http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
State-Changed-From-To: open->feedback
State-Changed-By: xtraeme@netbsd.org
State-Changed-When: Sun, 29 Jul 2007 15:19:38 +0000
State-Changed-Why:
- sme_mtx is not acquired anymore in the loop of sysmon_envsys_createplist().
- sysmon_envsys_register was changed to not drop the mutex before calling
sysmon_envsys_createplist().
Time to close it, or are there still more possible problems?
From: yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi)
To: juan@xtrarom.org
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Mon, 30 Jul 2007 07:53:19 +0900 (JST)
> On Wed, 25 Jul 2007 11:45:04 +0000 (UTC)
> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
>
> > can you provide a diff with -p?
>
> Sure, here it is: (I didn't know about -p, it's very useful :-)
can you explain what you are trying to achieve with this change?
it seems like almost no-op me. (except that it seems to introduce
a double-unlock bug of sme_list_mtx on some errors.)
YAMAMOTO Takashi
> Index: sysmon_envsys.c
> ===================================================================
> RCS file: /cvsroot/src/sys/dev/sysmon/sysmon_envsys.c,v
> retrieving revision 1.43
> diff -b -u -p -r1.43 sysmon_envsys.c
> --- sysmon_envsys.c 23 Jul 2007 17:51:16 -0000 1.43
> +++ sysmon_envsys.c 25 Jul 2007 12:34:15 -0000
> @@ -483,7 +483,6 @@ sysmon_envsys_register(struct sysmon_env
> mutex_enter(&sme_list_mtx);
> LIST_FOREACH(lsme, &sysmon_envsys_list, sme_list) {
> if (strcmp(lsme->sme_name, sme->sme_name) == 0) {
> - mutex_exit(&sme_list_mtx);
> error = EEXIST;
> goto out;
> }
> @@ -496,16 +495,14 @@ sysmon_envsys_register(struct sysmon_env
> sme->sme_fsensor = sysmon_envsys_next_sensor_index;
> sysmon_envsys_next_sensor_index += sme->sme_nsensors;
> #endif
> - mutex_exit(&sme_list_mtx);
> error = sysmon_envsys_createplist(sme);
> if (!error) {
> - mutex_enter(&sme_list_mtx);
> LIST_INSERT_HEAD(&sysmon_envsys_list, sme, sme_list);
> sme_uniqsensors = 0;
> - mutex_exit(&sme_list_mtx);
> }
>
> out:
> + mutex_exit(&sme_list_mtx);
> return error;
> }
>
> @@ -626,6 +623,9 @@ sysmon_envsys_createplist(struct sysmon_
>
> nsens = error = 0;
>
> + KASSERT(mutex_owned(&sme_list_mtx));
> + mutex_exit(&sme_list_mtx);
> +
> /* create the sysmon envsys device array. */
> array = prop_array_create();
> if (array == NULL)
> @@ -688,6 +688,7 @@ sysmon_envsys_createplist(struct sysmon_
>
>
> prop_object_release(array);
> + mutex_enter(&sme_list_mtx);
> return error;
> }
>
>
> --
> Juan Romero Pardines - The NetBSD Project
> http://plog.xtrarom.org - NetBSD/pkgsrc news in Spanish
State-Changed-From-To: feedback->open
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Sun, 04 May 2008 01:54:11 +0000
State-Changed-Why:
feedback received.
From: Juan RP <xtraeme@gmail.com>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/36673 (dubious code is sysmon_envsys)
Date: Sun, 4 May 2008 13:15:23 +0200
On Sun, 4 May 2008 01:54:12 +0000 (UTC)
dholland@NetBSD.org wrote:
> Synopsis: dubious code is sysmon_envsys
>
> State-Changed-From-To: feedback->open
> State-Changed-By: dholland@NetBSD.org
> State-Changed-When: Sun, 04 May 2008 01:54:11 +0000
> State-Changed-Why:
> feedback received.
Is this PR still valid? the code has been changed extensively
since this PR was sent and I think its points don't match anymore.
yamt, ok to close it or do you see any problem?
--
Juan Romero Pardines - xtraeme at gmail|netbsd dot org
The NetBSD Project
Make your own NetBSD/x86 Live CD
http://www.netbsd.org/~xtraeme/mklivecd/
Responsible-Changed-From-To: xtraeme->kern-bug-people
Responsible-Changed-By: wiz@NetBSD.org
Responsible-Changed-When: Sun, 25 May 2008 21:25:31 +0000
Responsible-Changed-Why:
Back to role account, xtraeme resigned.
State-Changed-From-To: open->feedback
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Mon, 25 Aug 2008 01:58:16 +0000
State-Changed-Why:
Is there anything that still needs attention here, or can this PR be closed?
State-Changed-From-To: feedback->closed
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Sat, 14 Nov 2009 23:38:56 +0000
State-Changed-Why:
>1 year feedback timeout.
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.