[added tech-kern to CC: as there seem to be some linux syscalls we lack]


On Tue, 19 Dec 2000, User Noud wrote:
>  12565 cfrdsservice RET   nanosleep 0
>  12565 cfrdsservice CALL  ipc(0x3,0xf0000,0,0x2,0xbfbfd44c)
>  12565 cfrdsservice RET   ipc 0  
>  12565 cfrdsservice CALL  nanosleep(0xbfbfd450,0)
>  12565 cfrdsservice RET   nanosleep 0
>  12565 cfrdsservice CALL  ipc(0x3,0xf0000,0,0x2,0xbfbfd44c)
>  12565 cfrdsservice RET   ipc 0
>  12565 cfrdsservice CALL  nanosleep(0xbfbfd450,0)
>  12565 cfrdsservice RET   nanosleep 0
>  12565 cfrdsservice CALL  write(0x2,0x4821e5a0,0x3c)
>  12565 cfrdsservice GIO   fd 2 wrote 60 bytes
>        "Error: performing semget() on ColdFusion Registry semaphore:"
>  12565 cfrdsservice RET   write 60/0x3c
>  12565 cfrdsservice CALL  write(0x2,0x483ba02a,0xb)  
>  12565 cfrdsservice GIO   fd 2 wrote 11 bytes
>        "File exists"  
>  12565 cfrdsservice RET   write 11/0xb
>  12565 cfrdsservice CALL  write(0x2,0x482a4067,0x1)
>  12565 cfrdsservice GIO   fd 2 wrote 1 bytes
>        "
>        "
>  12565 cfrdsservice RET   write 1
>  12565 cfrdsservice CALL  rt_sigprocmask(0x1,0xbfbfd350,0,0x8)
>  12565 cfrdsservice RET   rt_sigprocmask 0
>  12565 cfrdsservice CALL  kill(0x3115,0x6)
>  12565 cfrdsservice PSIG  SIGABRT SIG_DFL
>  12565 cfrdsservice NAMI  "cfrdsservice.core"

Seems like it's trying some operations on semaphores that fail. Do you
have options SYSVSEM in your kernel?


> i've got some:
> 
>  12561 cfexec   RET   dup2 1211912192/0x483c5000
>  12561 cfexec   CALL  #94 (unimplemented setdopt)(0x3,0x1a4)
>  12561 cfexec   RET   #94 (unimplemented setdopt) 0
>  12561 cfexec   CALL  write(0x3,0x483c5000,0x5)
> 
>  12562 cfexec   RET   read 148/0x94
>  12562 cfexec   CALL  #168 (unimplemented)(0x80d5600,0x1,0x7d0)
>  12562 cfexec   RET   #168 (unimplemented) 1
>  12562 cfexec   CALL  compat_43_ogetpagesize
> 
>  12561 cfexec   PSIG  SIGUSR1 caught handler=0x4822c000 mask=(1,2,15,20,30)
> code=0x0
>  12561 cfexec   RET   #179 (unimplemented) -1 errno -4 Unknown error: 4294967292
>  12561 cfexec   CALL  #119 (obsolete resuba)(0xbfbfd4f0)
>  12561 cfexec   RET   #119 (obsolete resuba) JUSTRETURN
>  12561 cfexec   CALL  ntp_gettime(0x2,0,0xbfbfd5b0,0x8)
> 
> are these unimplemented-thingys telling that
> i still miss some code? eg a function-call in a lib ??

No, that's system calls missing in the kernel. I'll CC: tech-kern on this
message, let's home someone there picks it up. ;)



> manually starting cfstdserver
> ./start-ing 4 cfexec's
> and after that manually start cfserver.
> 
> but that does dumps core too:
>  12757 cfserver NAMI  "/usr/pkg/emul/linux/opt/coldfusion/CustomTags/"
>  12757 cfserver RET   open 6
>  12757 cfserver CALL  compat_43_osigvec(0x6,0xbf1fc7a4)
>  12757 cfserver RET   compat_43_osigvec 0
>  12757 cfserver CALL  compat_12_oreboot(0x6,0x2,0x1) 
>  12757 cfserver RET   compat_12_oreboot 0 
>  12757 cfserver PSIG  SIGSEGV caught handler=0x489f5590 mask=(1,2,15,20,30)
> code=0x6
>  12757 cfserver NAMI  "cfserver.core"
>  12755 cfserver RET   kill 0
>  12755 cfserver CALL  #168 (unimplemented)(0x86ab100,0x1,0x7d0)
>  12754 cfserver GIO   fd 5 read 2265 bytes

Hum, left to our kernel gooroos, too.


> patching pkg-source and porting in is a thing to do for me,
> but this tracing is a bit to much for me.
> and i guess these trace snippets are to small to interpret.
> if any of you has some bright remark please do.
> think i have to lay-down the idee to get cf running on our suse-emul.
> for future ktrace's, i'd love to know the meaning of
> them "unimplemented", is it not found code ??
> thanx again & excuse my lengthy ktrace output.

Yeah, doesn't look too good right now.
What OS version are you running this on - 1.4.3, 1.5, -current?


 - Hubert

-- 
Hubert Feyrer <hubert@feyrer.de>