[linux-lvm] segfault running vgscan

[linux-lvm] segfault running vgscan

[Brad Langhorst]

I've carefully followed the instructions in INSTALL
and the build seems to go fine.
This is the first install of lvm so there are no vgs
or pvs or anything yet.

I patched the kernel  (2.4.9)
installed the kernel
make in LVM
make install
reboot
run vgscan -d  

this just generates a segfault and dumps core


any idea what i'm doing wrong?
nothing recent with vgscan and segfault in the
archive...
LVM version 1.0.1-rc1(17/08/2001) 
is reported in  dmesg.

thanks!


brad

__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Re: [linux-lvm] segfault running vgscan

[Heinz J . Mauelshagen]

Brad,

does it still segfault if you create a volume group before?

On Wed, Aug 22, 2001 at 06:10:53PM -0700, Brad Langhorst wrote:
> I've carefully followed the instructions in INSTALL
> and the build seems to go fine.
> This is the first install of lvm so there are no vgs
> or pvs or anything yet.
> 
> I patched the kernel  (2.4.9)
> installed the kernel
> make in LVM
> make install
> reboot
> run vgscan -d  
> 
> this just generates a segfault and dumps core
> 
> 
> any idea what i'm doing wrong?
> nothing recent with vgscan and segfault in the
> archive...
> LVM version 1.0.1-rc1(17/08/2001) 
> is reported in  dmesg.
> 
> thanks!
> 
> 
> brad
> 
> __________________________________________________
> Do You Yahoo!?
> Make international calls for as low as $.04/minute with Yahoo! Messenger
> http://phonecard.yahoo.com/
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] segfault running vgscan

[AJ Lewis]

[-- Attachment #1: Type: text/plain, Size: 1307 bytes --]

On Wed, Aug 22, 2001 at 06:10:53PM -0700, Brad Langhorst wrote:
> I've carefully followed the instructions in INSTALL
> and the build seems to go fine.
> This is the first install of lvm so there are no vgs
> or pvs or anything yet.
> 
> I patched the kernel  (2.4.9)
> installed the kernel
> make in LVM
> make install
> reboot
> run vgscan -d  
> 
> this just generates a segfault and dumps core

Are you linking the tools statically or dynamically?  I have seen a
seg-fault with the tools that call at_exit() and static linking with glibc -
apparently the bug is in glibc, not LVM, and is fixed in newer versions.  I
have yet to verify this.
 
-- 
AJ Lewis
Sistina Software Inc.                  Voice:  612-638-0500
1313 5th St SE, Suite 111              Fax:    612-638-0500
Minneapolis, MN 55414                  E-Mail: lewis@sistina.com
http://www.sistina.com

Current GPG fingerprint = 3B5F 6011 5216 76A5 2F6B  52A0 941E 1261 0029 2648
Get my key at: http://www.sistina.com/~lewis/gpgkey
 (Unfortunately, the PKS-type keyservers do not work with multiple sub-keys)

-----Begin Obligatory Humorous Quote----------------------------------------
All wiyht. Rho sritched mg kegtops awound?
-----End Obligatory Humorous Quote------------------------------------------

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

[linux-lvm] segfault running vgscan

[Brad Langhorst]

>Brad,
>
>does it still segfault if you create a volume group
>before?
>
>On Wed, Aug 22, 2001 at 06:10:53PM -0700, Brad
>Langhorst wrote:
>> I've carefully followed the instructions in INSTALL
>> and the build seems to go fine.
>>...
vgcreate also segfaults 
it seems clear to me that this is a build problem not
a code bug since almost everything segfaults

pvdisplay and pvscan do not segfault
all other pv commands and all vg commands segfault

brad


__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

[linux-lvm] segfault running vgscan

[Brad Langhorst]

>Are you linking the tools statically or dynamically?
> I have seen a
>seg-fault with the tools that call at_exit() and
>static linking with glibc -
>apparently the bug is in glibc, not LVM, and is
>fixed in newer versions.  I
>have yet to verify this.
>=20
>--=20
>AJ Lewis
sorry to be an idiot - i'm not sure how to tell if its
linked statically or dynamically.  Looking through the
build messags I see a gcc -shared command in the lib
building section so I'm guessing dynamic linking but I
don't really know.

here is my glibc info

# rpm -qa |grep glibc
compat-glibc-6.2-2.1.3.2
glibc-common-2.2.4-5
glibc-devel-2.2.4-5
glibc-2.2.4-5
glibc-profile-2.2.4-5

I ran the program using strace...
but I don't really know what everything means yet.
It looks to me like it is opening shared libraries and
allocating memory - but for some reason is not allowed
to allocate 0x40187000.
*shrug*

# strace vgscan
execve("/sbin/vgscan", ["vgscan"], [/* 27 vars */]) =
0
uname({sys="Linux", node="[hostname.removed]", ...}) =
0
brk(0)                                  = 0x804b3e0
open("/etc/ld.so.preload", O_RDONLY)    = -1 ENOENT
(No such file or directory)
open("/etc/ld.so.cache", O_RDONLY)      = 3
fstat64(3, {st_mode=S_IFREG|0644, st_size=38258, ...})
= 0
old_mmap(NULL, 38258, PROT_READ, MAP_PRIVATE, 3, 0) =
0x40017000
close(3)                                = 0
open("/lib/liblvm-10.so.1", O_RDONLY)   = 3
read(3,
"\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\360P\0"...,
1024) = 1024
fstat64(3, {st_mode=S_IFREG|0555, st_size=704519,
...}) = 0
old_mmap(NULL, 189300, PROT_READ|PROT_EXEC,
MAP_PRIVATE, 3, 0) = 0x40021000
mprotect(0x4004b000, 17268, PROT_NONE)  = 0
old_mmap(0x4004b000, 8192, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 3, 0x29000) = 0x4004b000
old_mmap(0x4004d000, 9076, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) =
0x4004d000
close(3)                                = 0
open("/lib/i686/libc.so.6", O_RDONLY)   = 3
read(3,
"\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\340\310"...,
1024) = 1024
fstat64(3, {st_mode=S_IFREG|0755, st_size=5761921,
...}) = 0
old_mmap(NULL, 4096, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40050000
old_mmap(NULL, 1285704, PROT_READ|PROT_EXEC,
MAP_PRIVATE, 3, 0) = 0x40051000
mprotect(0x40182000, 36424, PROT_NONE)  = 0
old_mmap(0x40182000, 20480, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 3, 0x130000) = 0x40182000
old_mmap(0x40187000, 15944, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) =
0x40187000
close(3)                                = 0
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++


__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Re: [linux-lvm] segfault running vgscan

[Mark J. Bobak]

Use the ldd command.  This 'list dynamic dependencies'.  If a program
has no dynamic dependencies, it's statically linked, by definition.

just 'ldd <name of binary>'.

-Mark

--
Mark J. Bobak
Oracle DBA
mark@bobak.net
"Blessed is the man who, having nothing to say, abstains from giving
 wordy evidence of the fact." -- George Eliot

On Thu, 23 Aug 2001, Brad Langhorst wrote:

>
> >Are you linking the tools statically or dynamically?
> > I have seen a
> >seg-fault with the tools that call at_exit() and
> >static linking with glibc -
> >apparently the bug is in glibc, not LVM, and is
> >fixed in newer versions.  I
> >have yet to verify this.
> >=20
> >--=20
> >AJ Lewis
> sorry to be an idiot - i'm not sure how to tell if its
> linked statically or dynamically.  Looking through the
> build messags I see a gcc -shared command in the lib
> building section so I'm guessing dynamic linking but I
> don't really know.
>
> here is my glibc info
>
> # rpm -qa |grep glibc
> compat-glibc-6.2-2.1.3.2
> glibc-common-2.2.4-5
> glibc-devel-2.2.4-5
> glibc-2.2.4-5
> glibc-profile-2.2.4-5
>
> I ran the program using strace...
> but I don't really know what everything means yet.
> It looks to me like it is opening shared libraries and
> allocating memory - but for some reason is not allowed
> to allocate 0x40187000.
> *shrug*
>
> # strace vgscan
> execve("/sbin/vgscan", ["vgscan"], [/* 27 vars */]) =
> 0
> uname({sys="Linux", node="[hostname.removed]", ...}) =
> 0
> brk(0)                                  = 0x804b3e0
> open("/etc/ld.so.preload", O_RDONLY)    = -1 ENOENT
> (No such file or directory)
> open("/etc/ld.so.cache", O_RDONLY)      = 3
> fstat64(3, {st_mode=S_IFREG|0644, st_size=38258, ...})
> = 0
> old_mmap(NULL, 38258, PROT_READ, MAP_PRIVATE, 3, 0) =
> 0x40017000
> close(3)                                = 0
> open("/lib/liblvm-10.so.1", O_RDONLY)   = 3
> read(3,
> "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\360P\0"...,
> 1024) = 1024
> fstat64(3, {st_mode=S_IFREG|0555, st_size=704519,
> ...}) = 0
> old_mmap(NULL, 189300, PROT_READ|PROT_EXEC,
> MAP_PRIVATE, 3, 0) = 0x40021000
> mprotect(0x4004b000, 17268, PROT_NONE)  = 0
> old_mmap(0x4004b000, 8192, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_FIXED, 3, 0x29000) = 0x4004b000
> old_mmap(0x4004d000, 9076, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) =
> 0x4004d000
> close(3)                                = 0
> open("/lib/i686/libc.so.6", O_RDONLY)   = 3
> read(3,
> "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\340\310"...,
> 1024) = 1024
> fstat64(3, {st_mode=S_IFREG|0755, st_size=5761921,
> ...}) = 0
> old_mmap(NULL, 4096, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40050000
> old_mmap(NULL, 1285704, PROT_READ|PROT_EXEC,
> MAP_PRIVATE, 3, 0) = 0x40051000
> mprotect(0x40182000, 36424, PROT_NONE)  = 0
> old_mmap(0x40182000, 20480, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_FIXED, 3, 0x130000) = 0x40182000
> old_mmap(0x40187000, 15944, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) =
> 0x40187000
> close(3)                                = 0
> --- SIGSEGV (Segmentation fault) ---
> +++ killed by SIGSEGV +++
>
>
> __________________________________________________
> Do You Yahoo!?
> Make international calls for as low as $.04/minute with Yahoo! Messenger
> http://phonecard.yahoo.com/
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html
>

Re: [linux-lvm] segfault running vgscan

[Brad Langhorst]

--- "Mark J. Bobak" <mark@bobak.net> wrote:
> Use the ldd command.  This 'list dynamic
> dependencies'.  If a program
> has no dynamic dependencies, it's statically linked,
> by definition.
> 
okay - ldd says that vgscan is linked to

#ldd vgscan
liblvm-10.so.1 => /lib/liblvm-10.so.1 (0x40021000)
libc.so.6 => /lib/i686/libc.so.6 (0x40051000)
/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)

so yes dynamically linked...
the problem was with static linking right?

now what? 
should I try a less optimized libc?

thanks 

brad

__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Re: [linux-lvm] segfault running vgscan

[AJ Lewis]

[-- Attachment #1: Type: text/plain, Size: 1706 bytes --]

On Thu, Aug 23, 2001 at 12:22:35PM -0700, Brad Langhorst wrote:
> so yes dynamically linked...
> the problem was with static linking right?
> 
> now what? 
> should I try a less optimized libc?

I know that there was an issue with atexit() with certain versions of
glibc, it's possibly both when dynamically and statically linked.  I'm not
sure.  I am sure that the following tools call atexit() through LVM_LOCK():

lvchange.c
lvcreate.c
lvdisplay.c
lvextend.c
lvm_user.h
lvm_user.h
lvmchange.c
lvmsadc.c
lvreduce.c
lvremove.c
lvrename.c
lvscan.c
pvchange.c
pvchange.c
pvcreate.c
pvdata.c
pvdisplay.c
pvmove.c
pvscan.c
vgcfgbackup.c
vgcfgrestore.c
vgchange.c
vgck.c
vgcreate.c
vgdisplay.c
vgexport.c
vgextend.c
vgimport.c
vgmerge.c
vgmknodes.c
vgreduce.c
vgremove.c
vgrename.c
vgscan.c
vgsplit.c

Can you verify that these are the ones that segfault, or that the other
tools do not?  Can you upgrade glibc?  Or try on a machine with a
different version?

Regards,
-- 
AJ Lewis
Sistina Software Inc.                  Voice:  612-638-0500
1313 5th St SE, Suite 111              Fax:    612-638-0500
Minneapolis, MN 55414                  E-Mail: lewis@sistina.com
http://www.sistina.com

Current GPG fingerprint = 3B5F 6011 5216 76A5 2F6B  52A0 941E 1261 0029 2648
Get my key at: http://www.sistina.com/~lewis/gpgkey
 (Unfortunately, the PKS-type keyservers do not work with multiple sub-keys)

-----Begin Obligatory Humorous Quote----------------------------------------
There's nary an animal alive that can outrun a greased Scotsman.
   - Groundskeeper Willie
-----End Obligatory Humorous Quote------------------------------------------

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] segfault running vgscan

[Brad Langhorst]

the static vs dynamic link bit gave me the idea to try
compiling lvm statically - seems to work now.

i can't upgrade glibc since its already at 2.2.4 which
is the most recent (as far as I can tell)


lets hope it doesn't just appear to work...

brad
--- AJ Lewis <lewis@sistina.com> wrote:
> On Thu, Aug 23, 2001 at 12:22:35PM -0700, Brad
> Langhorst wrote:
> > so yes dynamically linked...
> > the problem was with static linking right?
> > 
> > now what? 
> > should I try a less optimized libc?
> 
> I know that there was an issue with atexit() with
> certain versions of
> glibc, it's possibly both when dynamically and
> statically linked.  I'm not
> sure.  I am sure that the following tools call
> atexit() through LVM_LOCK():
> 
> lvchange.c
> lvcreate.c
> lvdisplay.c
> lvextend.c
> lvm_user.h
> lvm_user.h
> lvmchange.c
> lvmsadc.c
> lvreduce.c
> lvremove.c
> lvrename.c
> lvscan.c
> pvchange.c
> pvchange.c
> pvcreate.c
> pvdata.c
> pvdisplay.c
> pvmove.c
> pvscan.c
> vgcfgbackup.c
> vgcfgrestore.c
> vgchange.c
> vgck.c
> vgcreate.c
> vgdisplay.c
> vgexport.c
> vgextend.c
> vgimport.c
> vgmerge.c
> vgmknodes.c
> vgreduce.c
> vgremove.c
> vgrename.c
> vgscan.c
> vgsplit.c
> 
> Can you verify that these are the ones that
> segfault, or that the other
> tools do not?  Can you upgrade glibc?  Or try on a
> machine with a
> different version?
> 
> Regards,
> -- 
> AJ Lewis
> Sistina Software Inc.                  Voice: 
> 612-638-0500
> 1313 5th St SE, Suite 111              Fax:   
> 612-638-0500
> Minneapolis, MN 55414                  E-Mail:
> lewis@sistina.com
> http://www.sistina.com
> 
> Current GPG fingerprint = 3B5F 6011 5216 76A5 2F6B 
> 52A0 941E 1261 0029 2648
> Get my key at: http://www.sistina.com/~lewis/gpgkey
>  (Unfortunately, the PKS-type keyservers do not work
> with multiple sub-keys)
> 
> -----Begin Obligatory Humorous
> Quote----------------------------------------
> There's nary an animal alive that can outrun a
> greased Scotsman.
>    - Groundskeeper Willie
> -----End Obligatory Humorous
> Quote------------------------------------------
> 

> ATTACHMENT part 2 application/pgp-signature 


__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Re: [linux-lvm] segfault running vgscan

[AJ Lewis]

[-- Attachment #1: Type: text/plain, Size: 1226 bytes --]

On Thu, Aug 23, 2001 at 03:51:07PM -0700, Brad Langhorst wrote:
> the static vs dynamic link bit gave me the idea to try
> compiling lvm statically - seems to work now.
> 
> i can't upgrade glibc since its already at 2.2.4 which
> is the most recent (as far as I can tell)

Interesting that i have the exact opposite problem with version 2.2.3 of
glibc...I think more research is required.  Has anyone else seen a problem
like this?

> lets hope it doesn't just appear to work...

Yes!

-- 
AJ Lewis
Sistina Software Inc.                  Voice:  612-638-0500
1313 5th St SE, Suite 111              Fax:    612-638-0500
Minneapolis, MN 55414                  E-Mail: lewis@sistina.com
http://www.sistina.com

Current GPG fingerprint = 3B5F 6011 5216 76A5 2F6B  52A0 941E 1261 0029 2648
Get my key at: http://www.sistina.com/~lewis/gpgkey
 (Unfortunately, the PKS-type keyservers do not work with multiple sub-keys)

-----Begin Obligatory Humorous Quote----------------------------------------
"Facts are useless!  You can use facts to prove almost anything that's even
remotely true.  Facts schmacts."  --Homer Simpson
-----End Obligatory Humorous Quote------------------------------------------

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] segfault running vgscan

[Patrick Caulfield]

On Thu, Aug 23, 2001 at 07:54:02PM -0500, AJ Lewis wrote:
> On Thu, Aug 23, 2001 at 03:51:07PM -0700, Brad Langhorst wrote:
> > the static vs dynamic link bit gave me the idea to try
> > compiling lvm statically - seems to work now.
> > 
> > i can't upgrade glibc since its already at 2.2.4 which
> > is the most recent (as far as I can tell)
> 
> Interesting that i have the exact opposite problem with version 2.2.3 of
> glibc...I think more research is required.  Has anyone else seen a problem
> like this?

Just tried it with libc 2.2.4 and GCC 3.0 and GCC 2.95 and it seems OK for me.

patrick