Date: Thu, 12 Mar 1998 21:29:36 -0500 From: "Alfred Perlstein" <perlsta@cs.sunyit.edu> To: <hackers@FreeBSD.ORG> Subject: Re: SYSVMEM leak? thanks you but... Message-ID: <008001bd4e28$274f2ac0$0600a8c0@win95.local.sunyit.edu>
next in thread | raw e-mail | index | archive | help
i appreciate all the feedback from people on this issue, however it might be due to a buggy program here and there, however i'm quite sure it is due to a missing feature or bug in the Linux-EMU, i am interested in looking at this, but there is hardly anything that i can find to help me. the problem seems to manifest itself in StarOffice 4.0 and x11amp. they do not seem to be present in -stable, but they might, (i saw staroffice running for a bit on a -stable system) when i run staroffice it seems to run amok with allocating segments... it grabs a LOT of them.. in fact all of them, then it crashes before i can even do anything. it gives a large dump that ends in: SalImage::Create pShmInfo_->shmid < 0 (28) SalImage::Create pShmInfo_->shmid < 0 (28) then it just locks... some sort of runaway leakage of SYSVSHM in the linux emu. the problem occurs running x11amp, the linux libc version, it seems to lose 3 SYSVMEM segments everytime it playes a mp3... along with that a message is printed at the consol: Linux-emu(1024) - Setup() not implemented. the 1024 is the pid of the process, x11amp seems to be fine... but it dies when i finally run out of shared memory segments, i doubt this problem exists on linux boxes... i will rephrase my original e-mail: I know how to delete SYSVMEM segments by hand, but i don't want to have to do this... and no.... i do NOT want to run a cron job that deletes all my segments cause that is a bogus hack. what i need is: 1) a bit of understanding via a pointer to ANY sort of documentation on why i'm getting that error (obviously an unimplemented function in the linux emu) 2) why the emu is reporting this error (setup() not implemented)... i just checked on a friend's linux machine the man page for "setup()" and it seems that it is a function that is only called by the kernel only at bootup. i know linux documentation is horribly out of date.... and i do plan on mucking through the linux header files and whatnot for this... 3) a pointer and some adivice on where to look in the linux emu to examine this... 4) how to debug the x11amp (i'll figure this out on my own i'm sure) but i'm sure it is stripped. 5) are these segments being allocated on behalf of the program? or does the linux EMU allocate them to emulate some linux "feature"? thank you, -Alfred -----Original Message----- From: Joao Carlos Mendes Luis <jonny@coppe.ufrj.br> To: Alfred Perlstein <perlsta@cs.sunyit.edu> Cc: hackers@FreeBSD.ORG <hackers@FreeBSD.ORG> Date: Thursday, March 12, 1998 2:45 PM Subject: Re: SYSVMEM leak? >#define quoting(Alfred Perlstein) >// i've been playing with a bunch of programs and they all seem to like to >// leave around SYSVSHM segments, this is especially problematic in programs >// running under the linux emulation. >// >// i get a lot of messages on the consol: >// linux-emu(1040) - setup() not implemented. > >SYSSHM memory should be politely released by the program, as it's not >done at exit(). If the program has simply aborted, chances are that >the shmem is not being release. Take a look at ipcrm(1). > > Jonny > >-- >Joao Carlos Mendes Luis jonny@gta.ufrj.br >+55 21 290-4698 jonny@coppe.ufrj.br >Universidade Federal do Rio de Janeiro UFRJ/COPPE/CISI >PGP fingerprint: 29 C0 50 B9 B6 3E 58 F2 83 5F E3 26 BF 0F EA 67 > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?008001bd4e28$274f2ac0$0600a8c0>