Segmentation Fault using Gprint

[human]

rrdtool always seg fault when i try to use GPRINT.

Very simple example:

# rrdtool graph a.png DEF:a=a.rrd:cpu_nice:AVERAGE GPRINT:a:AVERAGE:%lf
Segmentation fault

# dmesg
Process rrdtool (pid: 16050, stack limit = 0xc2cc61a0)
Stack: (0xc2cc7fb0 to 0xc2cc8000)
7fa0: 00000000 000131b8 00032000 00011a70
7fc0: 00003dc9 00000000 00011a70 bed2cf2c 0000074c 00001c24 4006ba68 00da72ec
7fe0: 4006bba8 bed2cdc0 40036e00 40036ddc 20000010 ffffffff 00000000 00000000
Backtrace: no frame pointer
Code: e3a04000 e3a05102 e880003c eafffd28 (e5961000)

<1>Unable to handle kernel paging request at virtual address 00032000

pgd = c2654000
[00032000] *pgd=0428c031, *pte=00000000, *ppte=00000000
Internal error: Oops: 17 #9
Modules linked in:
CPU: 0
PC is at CPDT_load_double+0x0/0x44
LR is at fpundefinstr+0x0/0x20
pc : [<c004f7fc>] lr : [<c0024d60>] Not tainted
sp : c2cc7fb0 ip : 4006bba8 fp : 00000000
r10: c2cc6064 r9 : c0024dc0 r8 : 00000100
r7 : 00032000 r6 : 00032000 r5 : 00020000 r4 : ed92c100
r3 : 20000010 r2 : 40036ddc r1 : 00000005 r0 : c2cc60a4
Flags: nzcv IRQs on FIQs on Mode SVC_32 Segment user
Control: A005317F Table: 02654000 DAC: 00000015
Process rrdtool (pid: 16073, stack limit = 0xc2cc61a0)
Stack: (0xc2cc7fb0 to 0xc2cc8000)
7fa0: 00000000 000131b8 00032000 00011a70
7fc0: 00003dc9 00000000 00011a70 bedc5f2c 0000074c 00001c24 4006ba68 00da72ec
7fe0: 4006bba8 bedc5dc0 40036e00 40036ddc 20000010 ffffffff 00000000 00000000
Backtrace: no frame pointer
Code: e3a04000 e3a05102 e880003c eafffd28 (e5961000)

# uname -a
Linux LS-GL7D6 2.6.12.6-arm1 #77 Tue Dec 5 22:23:46 CST 2006 armv5tejl GNU/Linux

Any other info that i can provide ? strace is attached.

Right now i am using 1.2.15 from debian repository. But i have compiled 1.2.25 by myself and the problem remains.

[human]

stack trace

[oetiker]

There are tree possible reasons for this:

a) your rrd file in special in the sense that rrdtool can not deal with it ...

b) your kernel has issues with programs using mmap io

c) you do not have enough memory available ...

if you amend your bug report with a full demonstration script with create update graph commands, then this will be much easier to reproduce.

[oetiker]

closing the bug since I can not reporduce it without further information.