Re: [9fans] broken vac

Re: [9fans] broken vac

[Russ Cox]

did you do a pull while you were running that vac?
if so, it may have broken just because the text file
changed underneath it and it got confused when it
paged in a new section.

Re: [9fans] broken vac

[rog]

axel:
> I tried: cmp /proc/number/text /bin/vac
> which did not report differences; is that what I need,
> or will the text file point to the new binary if I update
> the binary on the system?

russ:
> The cmp is fine.   You have the latest vac.
> Something else is wrong.

actually, i'm not sure that's right.  the text file just points to the
file on the filesystem, so unless that's been removed/renamed in the
meantime (safeinstall?)  the cmp is unlikely to tell you much.
looking in the dump might be more informative.

  cheers,
    rog.

% cat tst.c
#include <u.h>
#include <libc.h>
void
main(void)
{
	sleep(2000000);
}
% 8c tst.c && 8l tst.8
% 8.out &
% echo $apid
600
% md5sum /proc/600/text
416079107feb6f50ad2204838f6a49c2	/proc/600/text
% ed tst.c
,s/2000/1000/p
	sleep(1000000);
w
70
q
% % 8c tst.c && 8l tst.8
% md5sum /proc/600/text
15d26908dcf5ee6a2919a255cd1f2a02	/proc/600/text
% md5sum 8.out
15d26908dcf5ee6a2919a255cd1f2a02	8.out
%

Re: [9fans] broken vac

[Axel Belinfante]

I am almost certain I did not.

I'll see if I can try to reproduce it, some time later.
I'm hesitant to try now since I still
have a vac running at the moment.
If I can reproduce I'll let you know.

FYI, possibly related:
is similar code in replica/applychanges (or similar name)?
Some weeks ago I tried to replica/pull to a pcmcia disk in a bitsy,
and the arm replica/applychanges broke. I did not want to spend
much time on it so I just exported the kfs from the bitsy to a pc
and did the replica/pull on the pc (which worked fine).
However, at the time I did have a quick look with acid on the
broken proces, and at least the function names seemed similar.

Axel.


> did you do a pull while you were running that vac?
> if so, it may have broken just because the text file
> changed underneath it and it got confused when it
> paged in a new section.

Re: [9fans] broken vac

[Russ Cox]

> FYI, possibly related:
> is similar code in replica/applychanges (or similar name)?

it's not a property of the program.  it's a property of
demand-paged executable loading in the kernel.

Re: [9fans] broken vac

[Fco.J.Ballesteros]

[-- Attachment #1: Type: text/plain, Size: 138 bytes --]

last time I checked, replica was taking care of running a binary
relocated to /tmp just to avoid the binary getting changed under feet.

[-- Attachment #2: Type: message/rfc822, Size: 3480 bytes --]

From: Axel Belinfante <Axel.Belinfante@cs.utwente.nl>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] broken vac
Date: Mon, 03 Feb 2003 16:36:49 +0100
Message-ID: <200302031536.h13FanR19682@zamenhof.cs.utwente.nl>

I am almost certain I did not.

I'll see if I can try to reproduce it, some time later.
I'm hesitant to try now since I still
have a vac running at the moment.
If I can reproduce I'll let you know.

FYI, possibly related:
is similar code in replica/applychanges (or similar name)?
Some weeks ago I tried to replica/pull to a pcmcia disk in a bitsy,
and the arm replica/applychanges broke. I did not want to spend
much time on it so I just exported the kfs from the bitsy to a pc
and did the replica/pull on the pc (which worked fine).
However, at the time I did have a quick look with acid on the
broken proces, and at least the function names seemed similar.

Axel.


> did you do a pull while you were running that vac?
> if so, it may have broken just because the text file
> changed underneath it and it got confused when it
> paged in a new section.

Re: [9fans] broken vac

[Russ Cox]

The cmp is fine.   You have the latest vac.
Something else is wrong.

Re: [9fans] broken vac

[Axel Belinfante]

> > If you have the most recent vac binary (Dec 15)
> > you should already have that copy of vtLockFree.

The broken vac still is around in broke(n) state.
Is there a way to check that it is indeed the Dec 15 version?
I tried: cmp /proc/number/text /bin/vac
which did not report differences; is that what I need,
or will the text file point to the new binary if I update
the binary on the system?

Axel.

Re: [9fans] broken vac

[Axel Belinfante]

> If you have the most recent vac binary (Dec 15)
> you should already have that copy of vtLockFree.

Hmmm...
I have that one, and I think that is the one I used.

In case it matters: the venti binary is from Dec 13
(With vac's running, I did not yet update venti).

Axel.

Re: [9fans] broken vac

[Russ Cox]

If you have the most recent vac binary (Dec 15)
you should already have that copy of vtLockFree.

Russ

Re: [9fans] broken vac

[Fco.J.Ballesteros]

Forgot to say: The same happen to me and they asked me to do that.
It now works fine.

Re: [9fans] broken vac

[Fco.J.Ballesteros]

[-- Attachment #1: Type: text/plain, Size: 192 bytes --]

Change vtLockFree with this:

void
vtLockFree(VtLock *p)
{
	if(p == nil)
		return;
	assert(p->writer == nil);
	assert(p->readers == 0);
	assert(p->qfirst == nil);
	vtMemFree(p);
}

[-- Attachment #2: Type: message/rfc822, Size: 5942 bytes --]

From: Axel Belinfante <Axel.Belinfante@cs.utwente.nl>
To: 9fans@cse.psu.edu
Subject: [9fans] broken vac
Date: Fri, 31 Jan 2003 16:08:46 +0100
Message-ID: <200301311508.h0VF8kD03966@zamenhof.cs.utwente.nl>

I'm trying to vac the dump of my current fs,
using a modified version of nemo's mkvac script.

At a particular point, this resulted in a broken vac,
see acid dump below. I retried (with same arguments),
with the same result.
Then I tried to vac other (more recent) /n/dump
directories, and that worked without problem.
Any ideas, anyone? If more information would help
to track down the problem, just let me know.

Axel.


The vac command was:

vac -s -f dump/ofs/2002_0101.vac -d dump/ofs/2003_0101.vac -q /n/dump/2002/0101

cpu% ls -l vac
--rwxrwxr-x M 1246 sys sys 173711 Dec 15 02:11 /386/bin/vac


cpu% acid 8720
/proc/8720/text:386 plan 9 executable

/sys/lib/acid/port
/sys/lib/acid/386
acid: lstk()
At pc:0x00010b2e:abort /sys/src/libc/9sys/abort.c:6
abort() /sys/src/libc/9sys/abort.c:6
	called from _assert+0x3a /sys/src/libc/port/_assert.c:12
_assert(s=0x0001b388) /sys/src/libc/port/_assert.c:7
	called from vtLockFree+0x54 /sys/src/libventi/plan9.c:245
vtLockFree(p=0x05090008) /sys/src/libventi/plan9.c:240
	called from vfFree+0x27 /sys/src/cmd/vac/file.c:265
vfFree(vf=0x0508ff48) /sys/src/cmd/vac/file.c:261
	called from vfDecRef+0x8f /sys/src/cmd/vac/file.c:983
vfDecRef(vf=0x0508ff48) /sys/src/cmd/vac/file.c:954
	called from vacDir+0x1cc /sys/src/cmd/vac/vac.c:608
	p=0x0506d968
	qq=0x0506d9f8
vacDir(dsink=0x0504ae60,fd=0x00000008,lname=0x0506d8e8,sname=0x0506d928,vf=0x05
06d968) /sys/src/cmd/vac/vac.c:581
	called from vacFile+0x17f /sys/src/cmd/vac/vac.c:437
	ds=0x0506dcc8
	dirs=0x0509ff08
	nd=0x00000043
	i=0x00000000
	name=0x050a0ebc
	ln=0x0508fea8
	sn=0x0508ff08
	vvf=0x0508ff48
vacFile(lname=0x0506d8e8,dsink=0x0504ae60,sname=0x0506d928,vf=0x0506d968) 
/sys/src/cmd/vac/vac.c:399
	called from vacDir+0x1bb /sys/src/cmd/vac/vac.c:605
	fd=0x00000008
	entry=0x00000000
	dir=0x0506dbe8
	vd=0x00000000
vacDir(dsink=0x050295d0,fd=0x00000007,lname=0x0504aac0,sname=0x0504ab00,vf=0x05
04ab40) /sys/src/cmd/vac/vac.c:581
	called from vacFile+0x17f /sys/src/cmd/vac/vac.c:437
	ds=0x0504ae60
	dirs=0x0507d948
	nd=0x00000009
	i=0x00000000
	name=0x0507db64
	ln=0x0506d8e8
	sn=0x0506d928
	vvf=0x0506d968
vacFile(lname=0x0504aac0,dsink=0x050295d0,sname=0x0504ab00,vf=0x0504ab40) 
/sys/src/cmd/vac/vac.c:399
	called from vacDir+0x1bb /sys/src/cmd/vac/vac.c:605
	fd=0x00000007
	entry=0x00000002
	dir=0x0504ad80
	vd=0x00000000
vacDir(dsink=0x00040438,fd=0x00000006,lname=0x7fffefea,sname=0x7fffeff7,vf=0x05
0292b0) /sys/src/cmd/vac/vac.c:581
	called from vacFile+0x17f /sys/src/cmd/vac/vac.c:437
	ds=0x050295d0
	dirs=0x0505ab68
	nd=0x0000001a
	i=0x00000001
	name=0x0505b1c1
	ln=0x0504aac0
	sn=0x0504ab00
	vvf=0x0504ab40
vacFile(lname=0x7fffefea,dsink=0x00040438,sname=0x7fffeff7,vf=0x050292b0) 
/sys/src/cmd/vac/vac.c:399
	called from vac+0x235 /sys/src/cmd/vac/vac.c:308
	fd=0x00000006
	entry=0x00000000
	dir=0x050294f0
	vd=0x00000000
vac(z=0x0001f328,argv=0x7fffefa0) /sys/src/cmd/vac/vac.c:253
	called from main+0x152 /sys/src/cmd/vac/vac.c:187
	cwd=0x7273752f
	dsink=0x00040438
	fs=0x00062488
	fd=0x00000004
	dir=0x050291d0
	cd=0x00000001
	cp2=0x7fffeff7
	vff=0x050292b0
	ms=0x000015c2
	vd=0x00000000
	ds=0x0000ccb0
	root=0x00012177
	buf=0x00000000
	score=0x00011f45
	i=0x00000000
main(argv=0x7fffefa0,argc=0x00000001) /sys/src/cmd/vac/vac.c:101
	called from _main+0x31 /sys/src/libc/386/main9.s:16
	host=0x00000000
	statsFlag=0x00000001
	_argc=0x00000071
	_args=0x7fffefe9
	p=0x00000000
	z=0x0001f328
acid: 

[9fans] broken vac

[Axel Belinfante]

I'm trying to vac the dump of my current fs,
using a modified version of nemo's mkvac script.

At a particular point, this resulted in a broken vac,
see acid dump below. I retried (with same arguments),
with the same result.
Then I tried to vac other (more recent) /n/dump
directories, and that worked without problem.
Any ideas, anyone? If more information would help
to track down the problem, just let me know.

Axel.


The vac command was:

vac -s -f dump/ofs/2002_0101.vac -d dump/ofs/2003_0101.vac -q /n/dump/2002/0101

cpu% ls -l vac
--rwxrwxr-x M 1246 sys sys 173711 Dec 15 02:11 /386/bin/vac


cpu% acid 8720
/proc/8720/text:386 plan 9 executable

/sys/lib/acid/port
/sys/lib/acid/386
acid: lstk()
At pc:0x00010b2e:abort /sys/src/libc/9sys/abort.c:6
abort() /sys/src/libc/9sys/abort.c:6
	called from _assert+0x3a /sys/src/libc/port/_assert.c:12
_assert(s=0x0001b388) /sys/src/libc/port/_assert.c:7
	called from vtLockFree+0x54 /sys/src/libventi/plan9.c:245
vtLockFree(p=0x05090008) /sys/src/libventi/plan9.c:240
	called from vfFree+0x27 /sys/src/cmd/vac/file.c:265
vfFree(vf=0x0508ff48) /sys/src/cmd/vac/file.c:261
	called from vfDecRef+0x8f /sys/src/cmd/vac/file.c:983
vfDecRef(vf=0x0508ff48) /sys/src/cmd/vac/file.c:954
	called from vacDir+0x1cc /sys/src/cmd/vac/vac.c:608
	p=0x0506d968
	qq=0x0506d9f8
vacDir(dsink=0x0504ae60,fd=0x00000008,lname=0x0506d8e8,sname=0x0506d928,vf=0x05
06d968) /sys/src/cmd/vac/vac.c:581
	called from vacFile+0x17f /sys/src/cmd/vac/vac.c:437
	ds=0x0506dcc8
	dirs=0x0509ff08
	nd=0x00000043
	i=0x00000000
	name=0x050a0ebc
	ln=0x0508fea8
	sn=0x0508ff08
	vvf=0x0508ff48
vacFile(lname=0x0506d8e8,dsink=0x0504ae60,sname=0x0506d928,vf=0x0506d968) 
/sys/src/cmd/vac/vac.c:399
	called from vacDir+0x1bb /sys/src/cmd/vac/vac.c:605
	fd=0x00000008
	entry=0x00000000
	dir=0x0506dbe8
	vd=0x00000000
vacDir(dsink=0x050295d0,fd=0x00000007,lname=0x0504aac0,sname=0x0504ab00,vf=0x05
04ab40) /sys/src/cmd/vac/vac.c:581
	called from vacFile+0x17f /sys/src/cmd/vac/vac.c:437
	ds=0x0504ae60
	dirs=0x0507d948
	nd=0x00000009
	i=0x00000000
	name=0x0507db64
	ln=0x0506d8e8
	sn=0x0506d928
	vvf=0x0506d968
vacFile(lname=0x0504aac0,dsink=0x050295d0,sname=0x0504ab00,vf=0x0504ab40) 
/sys/src/cmd/vac/vac.c:399
	called from vacDir+0x1bb /sys/src/cmd/vac/vac.c:605
	fd=0x00000007
	entry=0x00000002
	dir=0x0504ad80
	vd=0x00000000
vacDir(dsink=0x00040438,fd=0x00000006,lname=0x7fffefea,sname=0x7fffeff7,vf=0x05
0292b0) /sys/src/cmd/vac/vac.c:581
	called from vacFile+0x17f /sys/src/cmd/vac/vac.c:437
	ds=0x050295d0
	dirs=0x0505ab68
	nd=0x0000001a
	i=0x00000001
	name=0x0505b1c1
	ln=0x0504aac0
	sn=0x0504ab00
	vvf=0x0504ab40
vacFile(lname=0x7fffefea,dsink=0x00040438,sname=0x7fffeff7,vf=0x050292b0) 
/sys/src/cmd/vac/vac.c:399
	called from vac+0x235 /sys/src/cmd/vac/vac.c:308
	fd=0x00000006
	entry=0x00000000
	dir=0x050294f0
	vd=0x00000000
vac(z=0x0001f328,argv=0x7fffefa0) /sys/src/cmd/vac/vac.c:253
	called from main+0x152 /sys/src/cmd/vac/vac.c:187
	cwd=0x7273752f
	dsink=0x00040438
	fs=0x00062488
	fd=0x00000004
	dir=0x050291d0
	cd=0x00000001
	cp2=0x7fffeff7
	vff=0x050292b0
	ms=0x000015c2
	vd=0x00000000
	ds=0x0000ccb0
	root=0x00012177
	buf=0x00000000
	score=0x00011f45
	i=0x00000000
main(argv=0x7fffefa0,argc=0x00000001) /sys/src/cmd/vac/vac.c:101
	called from _main+0x31 /sys/src/libc/386/main9.s:16
	host=0x00000000
	statsFlag=0x00000001
	_argc=0x00000071
	_args=0x7fffefe9
	p=0x00000000
	z=0x0001f328
acid: