* [linux-lvm] pvmove
@ 2013-07-19 18:14 Martin Papik
2013-07-19 19:59 ` Zdenek Kabelac
0 siblings, 1 reply; 11+ messages in thread
From: Martin Papik @ 2013-07-19 18:14 UTC (permalink / raw)
To: linux-lvm
Hi everyone
I'm new to the mailing list but not to LVM, until now everything was
crystal clear and working fine. Now I'm either stuck or found a minor
issue. I'm trying to move move extents on a physical volume. Not from
one physical volume to another.
root@myhost# pvmove /dev/md127:151808-152063 /dev/md127:152064-152319
No extents available for allocation
root@myhost#
I also tried pvmove /dev/md127:151808-152063 /dev/md127:152064 and a few
other combination. There are enough free extents. The volume I've
created is 1G. There's 800 GB free. I made sure the destination blocks
are available, I've tried different ranges. Way below is the output of
vgdisplay and a list of volumes and their extents.
Is this functionality (extent shuffling) supported? I mean is this a
feature, bug or ignorance on my part?
PS, if you're wondering why would anyone want to move extents from a
device to the very same device, you may notice in the output below that
I have a few volumes spilt into pieces and I want to de-fragment it. I
know it won't affect performance too much, probably at all. But it is
going to help when (if?) I'm in dire need of dd if=/dev/xxx from the
physical volume. Wouldn't be the first time (not the fault of LVM,
disk+heat=bad blocks).
Thanks in advance for any ideas or scolding :-)
Martin
root@myhost# vgdisplay
--- Volume group ---
VG Name volume
System ID
Format lvm2
Metadata Areas 1
Metadata Sequence No 147
VG Access read/write
VG Status resizable
MAX LV 0
Cur LV 12
Open LV 9
Max PV 0
Cur PV 1
Act PV 1
VG Size 3.35 TiB
PE Size 4.00 MiB
Total PE 878905
Alloc PE / Size 666880 / 2.54 TiB
Free PE / Size 212025 / 828.22 GiB
VG UUID Gv2rC8-sMgQ-xGGK-5swv-9Gck-QA1G-k50ABy
root@myhost# vgs --all -o
+lv_name,lv_size,snap_percent,seg_pe_ranges,segtype,seg_size
VG #PV #LV #SN Attr VSize VFree LV LSize Snap% PE
Ranges Type SSize
volume 1 12 0 wz--n- 3.35t 828.22g home 150.00g
/dev/md127:0-25599 linear 100.00g
volume 1 12 0 wz--n- 3.35t 828.22g home 150.00g
/dev/md127:72085-74495 linear 9.42g
volume 1 12 0 wz--n- 3.35t 828.22g home 150.00g
/dev/md127:309248-319636 linear 40.58g
volume 1 12 0 wz--n- 3.35t 828.22g var 20.00g
/dev/md127:25600-30719 linear 20.00g
volume 1 12 0 wz--n- 3.35t 828.22g usr 20.00g
/dev/md127:30720-35839 linear 20.00g
volume 1 12 0 wz--n- 3.35t 828.22g tmp 2.00g
/dev/md127:35840-36351 linear 2.00g
volume 1 12 0 wz--n- 3.35t 828.22g p2p 300.00g
/dev/md127:74496-151295 linear 300.00g
volume 1 12 0 wz--n- 3.35t 828.22g boot 1.00g
/dev/md127:151296-151551 linear 1.00g
volume 1 12 0 wz--n- 3.35t 828.22g root 4.00g
/dev/md127:152320-153343 linear 4.00g
volume 1 12 0 wz--n- 3.35t 828.22g copy 2.00t
/dev/md127:181248-309247 linear 500.00g
volume 1 12 0 wz--n- 3.35t 828.22g copy 2.00t
/dev/md127:606208-712298 linear 414.42g
volume 1 12 0 wz--n- 3.35t 828.22g copy 2.00t
/dev/md127:156672-181247 linear 96.00g
volume 1 12 0 wz--n- 3.35t 828.22g copy 2.00t
/dev/md127:36352-72084 linear 139.58g
volume 1 12 0 wz--n- 3.35t 828.22g copy 2.00t
/dev/md127:319637-549524 linear 898.00g
volume 1 12 0 wz--n- 3.35t 828.22g swap 8.00g
/dev/md127:153344-155391 linear 8.00g
volume 1 12 0 wz--n- 3.35t 828.22g emulate 50.00g
/dev/md127:549525-562324 linear 50.00g
volume 1 12 0 wz--n- 3.35t 828.22g test 1.00g
/dev/md127:151552-151807 linear 1.00g
volume 1 12 0 wz--n- 3.35t 828.22g test2 1.00g
/dev/md127:151808-152063 linear 1.00g
root@myhost#
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-19 18:14 [linux-lvm] pvmove Martin Papik
@ 2013-07-19 19:59 ` Zdenek Kabelac
2013-07-19 22:27 ` Martin Papik
0 siblings, 1 reply; 11+ messages in thread
From: Zdenek Kabelac @ 2013-07-19 19:59 UTC (permalink / raw)
To: LVM general discussion and development; +Cc: Martin Papik
Dne 19.7.2013 20:14, Martin Papik napsal(a):
> Hi everyone
>
> I'm new to the mailing list but not to LVM, until now everything was crystal
> clear and working fine. Now I'm either stuck or found a minor issue. I'm
> trying to move move extents on a physical volume. Not from one physical volume
> to another.
>
> root@myhost# pvmove /dev/md127:151808-152063 /dev/md127:152064-152319
> No extents available for allocation
> root@myhost#
Add --alloc anywhere
But I assume the tool could be a bit more smarter here.
It looks more or less like a bug to me.
Zdenek
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-19 19:59 ` Zdenek Kabelac
@ 2013-07-19 22:27 ` Martin Papik
2013-07-19 23:20 ` Alasdair G Kergon
0 siblings, 1 reply; 11+ messages in thread
From: Martin Papik @ 2013-07-19 22:27 UTC (permalink / raw)
To: Zdenek Kabelac; +Cc: LVM general discussion and development
Dear Zdenek
Thank you very much, it works. I noticed a few more things. Not errors,
just minor things that might be documented better or handled
differently. Depending on the design philosophy.
"pvmove /dev/md127:152064-152319 /dev/md127:800000 --alloc anywhere"
fails because /dev/md127:xx seems to be interpreted as "one extent
starting at xx", I'm not sure that's what I'd guess from the man page.
"pvmove /dev/md127:152064-152319 /dev/md127:800000- --alloc anywhere"
works, but the manual page indicates that the syntax is
DestinationPhysicalVolume[:PE[-PE]...], which I would interpret as
/dev/xx:111 OR /dev/xx:111-222, but not /dev/xx:111-, that would IMHO be
DestinationPhysicalVolume[:PE[-[PE]]...].
I'd interpret /dev/md126:100 as a start of an area, since it has no
explicit bound, where as /dev/md127:100-100 I would expect for a single
extent, since there is an explicit bound. Does that make sense?
PS, how about a syntax variant where it can be /dev/md127:100+30 meaning
30 extents starting at 100? I understand it's often better to print
ranges, but maybe in some cases start+length might work better. Or
start-end+length for printing, and start[-end][+length] as input, and
exit with error if both end and length are specified and they don't
match (start+length!=length).
I'm not complaining, just suggesting enhancements. :-)
Again, thank you very much for your quick and helpful response.
Martin
On 07/19/2013 10:59 PM, Zdenek Kabelac wrote:
> Dne 19.7.2013 20:14, Martin Papik napsal(a):
>> Hi everyone
>>
>> I'm new to the mailing list but not to LVM, until now everything was
>> crystal
>> clear and working fine. Now I'm either stuck or found a minor issue. I'm
>> trying to move move extents on a physical volume. Not from one
>> physical volume
>> to another.
>>
>> root@myhost# pvmove /dev/md127:151808-152063 /dev/md127:152064-152319
>> No extents available for allocation
>> root@myhost#
>
> Add --alloc anywhere
>
> But I assume the tool could be a bit more smarter here.
> It looks more or less like a bug to me.
>
> Zdenek
>
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-19 22:27 ` Martin Papik
@ 2013-07-19 23:20 ` Alasdair G Kergon
2013-07-23 2:45 ` Martin Papik
0 siblings, 1 reply; 11+ messages in thread
From: Alasdair G Kergon @ 2013-07-19 23:20 UTC (permalink / raw)
To: Martin Papik; +Cc: LVM general discussion and development
It uses lists of ranges separated by colons.
:x-y
If x is missing, 0 is inserted; if y is missing, the last extent on the device
is inserted.
Man pages can always be improved with more examples!
For +length, try the patch below.
Alasdair
--- a/tools/toollib.c
+++ b/tools/toollib.c
@@ -1001,7 +1001,7 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
const char *pvname, uint32_t size)
{
char *endptr;
- uint32_t start, end;
+ uint32_t start, end, len;
/* Default to whole PV */
if (!c) {
@@ -1041,7 +1041,16 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
goto error;
c = endptr;
}
+ } else if (*c == '+') { /* Length? */
+ c++;
+ if (isdigit(*c)) {
+ if (!xstrtouint32(c, &endptr, 10, &len))
+ goto error;
+ c = endptr;
+ end = start + (len ? (len - 1) : 0);
+ }
}
+
if (*c && *c != ':')
goto error;
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-19 23:20 ` Alasdair G Kergon
@ 2013-07-23 2:45 ` Martin Papik
2013-07-24 3:33 ` [linux-lvm] pvmove ==> segfault: pvs --all -o +vg_fmt Martin Papik
` (3 more replies)
0 siblings, 4 replies; 11+ messages in thread
From: Martin Papik @ 2013-07-23 2:45 UTC (permalink / raw)
To: LVM general discussion and development
[-- Attachment #1: Type: text/plain, Size: 8000 bytes --]
Thank you. Is that patch going to be added to the main source code base?
I think I found an error (segfault), see below.
==> I have a few more improvement suggestions.
* IMHO vgs/lvs should include a column called seg_size_pe, similar to
seg_start and seg_start_pe
* IMHO vgs/lvs/pvs should have an command line parameter, or option
modifier allowing to not display empty columns, i.e. the following
command might skip displaying origin, snap%, move, log, copy%, convert,
because they have no values. That would allow me to show copy_percent
column only when copy is in progress. Would be nice.
lvs --all -o +vg_extent_size
LV VG Attr LSize Origin Snap% Move Log Copy% Convert Ext
boot volume -wi-ao 1.00g 4.00m
copy volume -wi-ao 2.00t 4.00m
emulate volume -wi-ao 50.00g 4.00m
home volume -wi-ao 150.00g 4.00m
p2p volume -wi-a- 300.00g 4.00m
root volume -wi-ao 4.00g 4.00m
swap volume -wi-ao 8.00g 4.00m
tmp volume -wi-ao 2.00g 4.00m
usr volume -wi-ao 20.00g 4.00m
var volume -wi-ao 20.00g 4.00m
Possible error:
root@localhost# lvs --rows --align
Does not align, should it?
*==> SEGFAULT INFO*
root@localhost# pvs --all -o +vg_fmt
Segmentation fault (core dumped)
However the following seems to work.....
root@localhost# pvs --all -o +vg_name
PV VG Fmt Attr PSize PFree VG
/dev/dm-0 -- 0 0
--- some entries removed for brevity ----
/dev/md127 volume lvm2 a- 5.46t 2.77t volume
--- some entries removed for brevity ----
/dev/sde1 -- 0 0
Additional information that might help.
root@localhost# pvs --version
LVM version: 2.02.66(2) (2010-05-20)
Library version: 1.02.48 (2010-05-20)
Driver version: 4.22.0
root@localhost# cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=12.04
DISTRIB_CODENAME=precise
DISTRIB_DESCRIPTION="Ubuntu 12.04.2 LTS"
root@localhost#valgrind pvs --all -o +vg_fmt
==14584== Memcheck, a memory error detector
==14584== Copyright (C) 2002-2011, and GNU GPL'd, by Julian Seward et al.
==14584== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright info
==14584== Command: pvs --all -o +vg_fmt
==14584==
==14584== Invalid read of size 8
==14584== at 0x548ED77: dm_report_field_string (in
/lib/libdevmapper.so.1.02.1)
==14584== by 0x549006F: dm_report_object (in /lib/libdevmapper.so.1.02.1)
==14584== by 0x46B138: report_object (in /sbin/lvm)
==14584== by 0x423618: ??? (in /sbin/lvm)
==14584== by 0x4254B2: process_each_pv (in /sbin/lvm)
==14584== by 0x4230D3: ??? (in /sbin/lvm)
==14584== by 0x41AE60: lvm_run_command (in /sbin/lvm)
==14584== by 0x41B307: lvm2_main (in /sbin/lvm)
==14584== by 0x58CE76C: (below main) (libc-start.c:226)
==14584== Address 0x28 is not stack'd, malloc'd or (recently) free'd
==14584==
==14584==
==14584== Process terminating with default action of signal 11 (SIGSEGV)
==14584== Access not within mapped region at address 0x28
==14584== at 0x548ED77: dm_report_field_string (in
/lib/libdevmapper.so.1.02.1)
==14584== by 0x549006F: dm_report_object (in /lib/libdevmapper.so.1.02.1)
==14584== by 0x46B138: report_object (in /sbin/lvm)
==14584== by 0x423618: ??? (in /sbin/lvm)
==14584== by 0x4254B2: process_each_pv (in /sbin/lvm)
==14584== by 0x4230D3: ??? (in /sbin/lvm)
==14584== by 0x41AE60: lvm_run_command (in /sbin/lvm)
==14584== by 0x41B307: lvm2_main (in /sbin/lvm)
==14584== by 0x58CE76C: (below main) (libc-start.c:226)
==14584== If you believe this happened as a result of a stack
==14584== overflow in your program's main thread (unlikely but
==14584== possible), you can try to increase the size of the
==14584== main thread stack using the --main-stacksize= flag.
==14584== The main thread stack size used in this run was 8388608.
==14584==
root@localhost# ltrace pvs --all -o +vg_fmt 2>&1 | tail
__ctype_toupper_loc() = 0x7f2f0e83f768
__ctype_toupper_loc() = 0x7f2f0e83f768
__ctype_toupper_loc() = 0x7f2f0e83f768
__sprintf_chk(0x1e86e88, 1, -1, 0x48daa5, 0x48dab8) = 2
dm_pool_strdup(0x1e88120, 0x1e86e88, 0x1e86e8a, 0, 0x48daa8) = 0x1e8e538
dm_pool_alloc(0x1e88120, 8, 2, 0x7f2f0d892030, 0x48daa8) = 0x1e8e540
dm_report_field_set_value(0x1e8e508, 0x1e8e538, 0x1e8e540,
0x7f2f0d892030, 0x48daa8) = 0x1e8e540
dm_report_field_string(0x1e88080, 0x1e8e550, 40, 0x6b1fe0, 0x1e6cfe0
<unfinished ...>
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++
root@localhost#
root@localhost# gdb --args pvs --all -o +vg_fmt
GNU gdb (Ubuntu/Linaro 7.4-2012.04-0ubuntu2.1) 7.4-2012.04
Copyright (C) 2012 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later
<http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
For bug reporting instructions, please see:
<http://bugs.launchpad.net/gdb-linaro/>...
Reading symbols from /sbin/pvs...(no debugging symbols found)...done.
(gdb) run
Starting program: /sbin/pvs --all -o +vg_fmt
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
File descriptor 3 (socket:[46992174]) leaked on pvs invocation. Parent
PID 14623: gdb
File descriptor 4 (socket:[46992175]) leaked on pvs invocation. Parent
PID 14623: gdb
File descriptor 5 (pipe:[46992176]) leaked on pvs invocation. Parent PID
14623: gdb
File descriptor 6 (pipe:[46992176]) leaked on pvs invocation. Parent PID
14623: gdb
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff757dd77 in dm_report_field_string () from
/lib/libdevmapper.so.1.02.1
(gdb) disassemble dm_report_field_string
Dump of assembler code for function dm_report_field_string:
0x00007ffff757dd70 <+0>: push %rbx
0x00007ffff757dd71 <+1>: mov %rsi,%rbx
0x00007ffff757dd74 <+4>: mov (%rdi),%rdi
=> 0x00007ffff757dd77 <+7>: mov (%rdx),%rsi
0x00007ffff757dd7a <+10>: callq 0x7ffff7570a30 <dm_pool_strdup@plt>
0x00007ffff757dd7f <+15>: test %rax,%rax
(gdb) info registers
rax 0x6dd550 7198032
rbx 0x6dd550 7198032
rcx 0x6b1fe0 7020512
rdx 0x28 40
rsi 0x6dd550 7198032
rdi 0x6d7120 7172384
rbp 0x6d7080 0x6d7080
On 07/20/2013 02:20 AM, Alasdair G Kergon wrote:
> It uses lists of ranges separated by colons.
>
> :x-y
>
> If x is missing, 0 is inserted; if y is missing, the last extent on the device
> is inserted.
>
> Man pages can always be improved with more examples!
>
> For +length, try the patch below.
>
> Alasdair
>
>
> --- a/tools/toollib.c
> +++ b/tools/toollib.c
> @@ -1001,7 +1001,7 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
> const char *pvname, uint32_t size)
> {
> char *endptr;
> - uint32_t start, end;
> + uint32_t start, end, len;
>
> /* Default to whole PV */
> if (!c) {
> @@ -1041,7 +1041,16 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
> goto error;
> c = endptr;
> }
> + } else if (*c == '+') { /* Length? */
> + c++;
> + if (isdigit(*c)) {
> + if (!xstrtouint32(c, &endptr, 10, &len))
> + goto error;
> + c = endptr;
> + end = start + (len ? (len - 1) : 0);
> + }
> }
> +
> if (*c && *c != ':')
> goto error;
>
[-- Attachment #2: Type: text/html, Size: 16750 bytes --]
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove ==> segfault: pvs --all -o +vg_fmt
2013-07-23 2:45 ` Martin Papik
@ 2013-07-24 3:33 ` Martin Papik
2013-09-23 20:38 ` [linux-lvm] pvmove Alasdair G Kergon
` (2 subsequent siblings)
3 siblings, 0 replies; 11+ messages in thread
From: Martin Papik @ 2013-07-24 3:33 UTC (permalink / raw)
To: LVM general discussion and development
[-- Attachment #1: Type: text/plain, Size: 8465 bytes --]
No reply yet so I thought I'd modify the subject to catch someone's eye.
Martin
On 07/23/2013 05:45 AM, Martin Papik wrote:
>
> Thank you. Is that patch going to be added to the main source code base?
>
> I think I found an error (segfault), see below.
>
> ==> I have a few more improvement suggestions.
>
> * IMHO vgs/lvs should include a column called seg_size_pe, similar to
> seg_start and seg_start_pe
> * IMHO vgs/lvs/pvs should have an command line parameter, or option
> modifier allowing to not display empty columns, i.e. the following
> command might skip displaying origin, snap%, move, log, copy%,
> convert, because they have no values. That would allow me to show
> copy_percent column only when copy is in progress. Would be nice.
>
> lvs --all -o +vg_extent_size
> LV VG Attr LSize Origin Snap% Move Log Copy% Convert Ext
> boot volume -wi-ao 1.00g 4.00m
> copy volume -wi-ao 2.00t 4.00m
> emulate volume -wi-ao 50.00g 4.00m
> home volume -wi-ao 150.00g
> 4.00m
> p2p volume -wi-a- 300.00g
> 4.00m
> root volume -wi-ao 4.00g 4.00m
> swap volume -wi-ao 8.00g 4.00m
> tmp volume -wi-ao 2.00g 4.00m
> usr volume -wi-ao 20.00g 4.00m
> var volume -wi-ao 20.00g 4.00m
>
> Possible error:
>
> root@localhost# lvs --rows --align
> Does not align, should it?
>
> *==> SEGFAULT INFO*
>
> root@localhost# pvs --all -o +vg_fmt
> Segmentation fault (core dumped)
>
> However the following seems to work.....
>
> root@localhost# pvs --all -o +vg_name
> PV VG Fmt Attr PSize PFree VG
> /dev/dm-0 -- 0 0
> --- some entries removed for brevity ----
> /dev/md127 volume lvm2 a- 5.46t 2.77t volume
> --- some entries removed for brevity ----
> /dev/sde1 -- 0 0
>
> Additional information that might help.
>
> root@localhost# pvs --version
> LVM version: 2.02.66(2) (2010-05-20)
> Library version: 1.02.48 (2010-05-20)
> Driver version: 4.22.0
> root@localhost# cat /etc/lsb-release
> DISTRIB_ID=Ubuntu
> DISTRIB_RELEASE=12.04
> DISTRIB_CODENAME=precise
> DISTRIB_DESCRIPTION="Ubuntu 12.04.2 LTS"
>
> root@localhost#valgrind pvs --all -o +vg_fmt
> ==14584== Memcheck, a memory error detector
> ==14584== Copyright (C) 2002-2011, and GNU GPL'd, by Julian Seward et al.
> ==14584== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright
> info
> ==14584== Command: pvs --all -o +vg_fmt
> ==14584==
> ==14584== Invalid read of size 8
> ==14584== at 0x548ED77: dm_report_field_string (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x549006F: dm_report_object (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x46B138: report_object (in /sbin/lvm)
> ==14584== by 0x423618: ??? (in /sbin/lvm)
> ==14584== by 0x4254B2: process_each_pv (in /sbin/lvm)
> ==14584== by 0x4230D3: ??? (in /sbin/lvm)
> ==14584== by 0x41AE60: lvm_run_command (in /sbin/lvm)
> ==14584== by 0x41B307: lvm2_main (in /sbin/lvm)
> ==14584== by 0x58CE76C: (below main) (libc-start.c:226)
> ==14584== Address 0x28 is not stack'd, malloc'd or (recently) free'd
> ==14584==
> ==14584==
> ==14584== Process terminating with default action of signal 11 (SIGSEGV)
> ==14584== Access not within mapped region at address 0x28
> ==14584== at 0x548ED77: dm_report_field_string (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x549006F: dm_report_object (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x46B138: report_object (in /sbin/lvm)
> ==14584== by 0x423618: ??? (in /sbin/lvm)
> ==14584== by 0x4254B2: process_each_pv (in /sbin/lvm)
> ==14584== by 0x4230D3: ??? (in /sbin/lvm)
> ==14584== by 0x41AE60: lvm_run_command (in /sbin/lvm)
> ==14584== by 0x41B307: lvm2_main (in /sbin/lvm)
> ==14584== by 0x58CE76C: (below main) (libc-start.c:226)
> ==14584== If you believe this happened as a result of a stack
> ==14584== overflow in your program's main thread (unlikely but
> ==14584== possible), you can try to increase the size of the
> ==14584== main thread stack using the --main-stacksize= flag.
> ==14584== The main thread stack size used in this run was 8388608.
> ==14584==
>
> root@localhost# ltrace pvs --all -o +vg_fmt 2>&1 | tail
> __ctype_toupper_loc() = 0x7f2f0e83f768
> __ctype_toupper_loc() = 0x7f2f0e83f768
> __ctype_toupper_loc() = 0x7f2f0e83f768
> __sprintf_chk(0x1e86e88, 1, -1, 0x48daa5, 0x48dab8) = 2
> dm_pool_strdup(0x1e88120, 0x1e86e88, 0x1e86e8a, 0, 0x48daa8) = 0x1e8e538
> dm_pool_alloc(0x1e88120, 8, 2, 0x7f2f0d892030, 0x48daa8) = 0x1e8e540
> dm_report_field_set_value(0x1e8e508, 0x1e8e538, 0x1e8e540,
> 0x7f2f0d892030, 0x48daa8) = 0x1e8e540
> dm_report_field_string(0x1e88080, 0x1e8e550, 40, 0x6b1fe0, 0x1e6cfe0
> <unfinished ...>
> --- SIGSEGV (Segmentation fault) ---
> +++ killed by SIGSEGV +++
> root@localhost#
>
> root@localhost# gdb --args pvs --all -o +vg_fmt
> GNU gdb (Ubuntu/Linaro 7.4-2012.04-0ubuntu2.1) 7.4-2012.04
> Copyright (C) 2012 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later
> <http://gnu.org/licenses/gpl.html>
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law. Type "show copying"
> and "show warranty" for details.
> This GDB was configured as "x86_64-linux-gnu".
> For bug reporting instructions, please see:
> <http://bugs.launchpad.net/gdb-linaro/>...
> Reading symbols from /sbin/pvs...(no debugging symbols found)...done.
> (gdb) run
> Starting program: /sbin/pvs --all -o +vg_fmt
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
> File descriptor 3 (socket:[46992174]) leaked on pvs invocation. Parent
> PID 14623: gdb
> File descriptor 4 (socket:[46992175]) leaked on pvs invocation. Parent
> PID 14623: gdb
> File descriptor 5 (pipe:[46992176]) leaked on pvs invocation. Parent
> PID 14623: gdb
> File descriptor 6 (pipe:[46992176]) leaked on pvs invocation. Parent
> PID 14623: gdb
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x00007ffff757dd77 in dm_report_field_string () from
> /lib/libdevmapper.so.1.02.1
> (gdb) disassemble dm_report_field_string
> Dump of assembler code for function dm_report_field_string:
> 0x00007ffff757dd70 <+0>: push %rbx
> 0x00007ffff757dd71 <+1>: mov %rsi,%rbx
> 0x00007ffff757dd74 <+4>: mov (%rdi),%rdi
> => 0x00007ffff757dd77 <+7>: mov (%rdx),%rsi
> 0x00007ffff757dd7a <+10>: callq 0x7ffff7570a30 <dm_pool_strdup@plt>
> 0x00007ffff757dd7f <+15>: test %rax,%rax
> (gdb) info registers
> rax 0x6dd550 7198032
> rbx 0x6dd550 7198032
> rcx 0x6b1fe0 7020512
> rdx 0x28 40
> rsi 0x6dd550 7198032
> rdi 0x6d7120 7172384
> rbp 0x6d7080 0x6d7080
>
>
>
> On 07/20/2013 02:20 AM, Alasdair G Kergon wrote:
>> It uses lists of ranges separated by colons.
>>
>> :x-y
>>
>> If x is missing, 0 is inserted; if y is missing, the last extent on the device
>> is inserted.
>>
>> Man pages can always be improved with more examples!
>>
>> For +length, try the patch below.
>>
>> Alasdair
>>
>>
>> --- a/tools/toollib.c
>> +++ b/tools/toollib.c
>> @@ -1001,7 +1001,7 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
>> const char *pvname, uint32_t size)
>> {
>> char *endptr;
>> - uint32_t start, end;
>> + uint32_t start, end, len;
>>
>> /* Default to whole PV */
>> if (!c) {
>> @@ -1041,7 +1041,16 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
>> goto error;
>> c = endptr;
>> }
>> + } else if (*c == '+') { /* Length? */
>> + c++;
>> + if (isdigit(*c)) {
>> + if (!xstrtouint32(c, &endptr, 10, &len))
>> + goto error;
>> + c = endptr;
>> + end = start + (len ? (len - 1) : 0);
>> + }
>> }
>> +
>> if (*c && *c != ':')
>> goto error;
>>
>
[-- Attachment #2: Type: text/html, Size: 17693 bytes --]
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-23 2:45 ` Martin Papik
2013-07-24 3:33 ` [linux-lvm] pvmove ==> segfault: pvs --all -o +vg_fmt Martin Papik
@ 2013-09-23 20:38 ` Alasdair G Kergon
2013-09-23 20:56 ` Alasdair G Kergon
2013-09-23 21:10 ` Alasdair G Kergon
3 siblings, 0 replies; 11+ messages in thread
From: Alasdair G Kergon @ 2013-09-23 20:38 UTC (permalink / raw)
To: Martin Papik; +Cc: LVM general discussion and development
On Tue, Jul 23, 2013 at 05:45:50AM +0300, Martin Papik wrote:
> Thank you. Is that patch going to be added to the main source code base?
Committed for 2.02.103.
> I think I found an error (segfault), see below.
> LVM version: 2.02.66(2) (2010-05-20)
I've no interest in bugs reported against such an old version of the code, I'm
afraid. If you reproduce with upstream code, then we'll look into it.
Alasdair
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-23 2:45 ` Martin Papik
2013-07-24 3:33 ` [linux-lvm] pvmove ==> segfault: pvs --all -o +vg_fmt Martin Papik
2013-09-23 20:38 ` [linux-lvm] pvmove Alasdair G Kergon
@ 2013-09-23 20:56 ` Alasdair G Kergon
2013-09-23 21:10 ` Alasdair G Kergon
3 siblings, 0 replies; 11+ messages in thread
From: Alasdair G Kergon @ 2013-09-23 20:56 UTC (permalink / raw)
To: Martin Papik; +Cc: LVM general discussion and development
On Tue, Jul 23, 2013 at 05:45:50AM +0300, Martin Papik wrote:
> ==> I have a few more improvement suggestions.
> * IMHO vgs/lvs should include a column called seg_size_pe, similar to
> seg_start and seg_start_pe
Implemented for 2.02.103.
https://git.fedorahosted.org/cgit/lvm2.git/commit/?id=11dc6a03c4cdd10bc4065c7d7a8585373455f947
Alasdair
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2013-07-23 2:45 ` Martin Papik
` (2 preceding siblings ...)
2013-09-23 20:56 ` Alasdair G Kergon
@ 2013-09-23 21:10 ` Alasdair G Kergon
3 siblings, 0 replies; 11+ messages in thread
From: Alasdair G Kergon @ 2013-09-23 21:10 UTC (permalink / raw)
To: Martin Papik; +Cc: LVM general discussion and development
On Tue, Jul 23, 2013 at 05:45:50AM +0300, Martin Papik wrote:
> * IMHO vgs/lvs/pvs should have an command line parameter, or option
> modifier allowing to not display empty columns, i.e. the following
> command might skip displaying origin, snap%, move, log, copy%, convert,
> because they have no values. That would allow me to show copy_percent
> column only when copy is in progress. Would be nice.
Yes - I'm surprised we never had this suggested before.
- Some columns will gain a property saying they can be suppressed when empty.
- A new cmdline option/config setting to suppress empty fields.
- Implemented as an extension alongside the 'sort' option. (Like sort, you
have to have all the output data prepared before you find out whether any
columns are empty - even the headings can't be output before this point.)
- A way to flag an output data entry as empty. (zkabelac is about to commit
some patches to tidy up the handling of empty fields so this change can
follow those.)
Alasdair
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [linux-lvm] pvmove
2003-07-12 7:44 Juan Pablo Giménez
@ 2003-07-12 8:29 ` Joe Thornber
0 siblings, 0 replies; 11+ messages in thread
From: Joe Thornber @ 2003-07-12 8:29 UTC (permalink / raw)
To: linux-lvm
On Fri, Jul 11, 2003 at 07:12:26PM -0300, Juan Pablo Gim?nez wrote:
> What's more secure, the Redhat patched pvmove or the original one?
The origininal pvmove has some serious problems that can make it
deadlock. Redhats version attempts to correct these.
The most recent LVM2 tools (see Alasdairs recent email) contain a
pvmove implementation as well.
- Joe
^ permalink raw reply [flat|nested] 11+ messages in thread
* [linux-lvm] pvmove
@ 2003-07-12 7:44 Juan Pablo Giménez
2003-07-12 8:29 ` Joe Thornber
0 siblings, 1 reply; 11+ messages in thread
From: Juan Pablo Giménez @ 2003-07-12 7:44 UTC (permalink / raw)
To: linux-lvm
[-- Attachment #1.1: Type: text/plain, Size: 897 bytes --]
What's more secure, the Redhat patched pvmove or the original one?
BTW, I have a little patch to fix RedHat patched pvmove and make it work
with unpatched kernels...
--- lib/pv_move.c 2003-07-11 19:10:26.000000000 -0300
+++ lib/pv_move.c.rcom 2003-07-11 19:09:35.000000000 -0300
@@ -916,7 +916,7 @@
ret = ioctl(group, PE_LOCKED_COPY, &pe_copy_req);
if (ret < 0)
ret = -errno;
- if (ret == -EINVAL)
+ if (ret == -EINVAL || ret == -ENOTTY)
ret = -LVM_EPV_LOCKED_COPY_EINVAL;
close(group);
Juan Pablo Giménez
http://www.rcom.com.ar/
http://www.rcom.com.ar/comunidad/
Llave GnuPG:
MIT keyserver
Web Rcom
GnuPG fingerprint: ED97 6FB8 A9BE
26A6 789B 6657 0A15 EC2B D3D8 FD2B
http://www.greenpeace.org.ar/secciones/cyberactivismo/nuclear.php3
[-- Attachment #1.2: Type: text/html, Size: 2027 bytes --]
[-- Attachment #2: Esta parte del mensaje está firmada digitalmente --]
[-- Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2013-09-23 21:10 UTC | newest]
Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-07-19 18:14 [linux-lvm] pvmove Martin Papik
2013-07-19 19:59 ` Zdenek Kabelac
2013-07-19 22:27 ` Martin Papik
2013-07-19 23:20 ` Alasdair G Kergon
2013-07-23 2:45 ` Martin Papik
2013-07-24 3:33 ` [linux-lvm] pvmove ==> segfault: pvs --all -o +vg_fmt Martin Papik
2013-09-23 20:38 ` [linux-lvm] pvmove Alasdair G Kergon
2013-09-23 20:56 ` Alasdair G Kergon
2013-09-23 21:10 ` Alasdair G Kergon
-- strict thread matches above, loose matches on Subject: below --
2003-07-12 7:44 Juan Pablo Giménez
2003-07-12 8:29 ` Joe Thornber
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.