Date: Fri, 23 Jun 2000 23:13:16 +0300 (EEST) From: "Jukka A. Ukkonen" <jau@cs94004.pp.htv.fi> To: FreeBSD-gnats-submit@freebsd.org Subject: i386/19477: linux_rt_sigprocmask() raises SIGSYS Message-ID: <200006232013.XAA67061@cs94004.pp.htv.fi>
next in thread | raw e-mail | index | archive | help
>Number: 19477 >Category: i386 >Synopsis: linux_rt_sigprocmask() raises SIGSYS >Confidential: no >Severity: critical >Priority: high >Responsible: freebsd-bugs >State: open >Quarter: >Keywords: >Date-Required: >Class: sw-bug >Submitter-Id: current-users >Arrival-Date: Fri Jun 23 13:20:00 PDT 2000 >Closed-Date: >Last-Modified: >Originator: Jukka A. Ukkonen >Release: FreeBSD 3.4-RELEASE i386 >Organization: Private person >Environment: FreeBSD 3.4-RELEASE on i386 with the Linux system calls emulation kernel module loaded >Description: Any Linux program that calls linux_rt_sigprocmask gets SIGSYS and dies on FreeBSD-3.4. Quite a number of Linux binaries seem to make such a call. When running FreeBSD-3.2 the same binaries worked just fine. >How-To-Repeat: Take e.g. the FrameMaker for Linux which is now distributed as a free beta test version. It is the latest Linux program I have had problems with. It can be downloaded from www.adobe.com. (You need to modify the start scripts a bit to make it start normally.) Once the start scripts are fixed to recognize FreeBSD as just another form of "Linux" add ktrace to the script which tries to start the real binary, and start the maker script. The binary crashes immediately at the beginning of the program. Run linux_kdump and look for a call to linux_rt_sigprogmask and SIGSYS at the end of the dump. Several, practically all Linux binaries, except some rather old ones, I have tried seem to produce the same result. Luckily I neither need nor use many Linux binaries though. I gave this report critical priority because this sort of problem could make the whole Linux emulation useless if many Linux programs are using the rt_sigprocmask() system call. As much as I know such a call could be part of the crt0.o for some or even all relatively new Linux versions. This seems to happen so early in the program that it might not have even reached the main() function yet. >Fix: I have not found a fix yet. In fact I guess there might have been a reason for someone changing the behaviour of linux system calls this way, but if there is such a reason I would sure like to know why. I tried to asking people for known reasons or known bugs for this odd behaviour, but apparently nobody has been able to give me any explanation. >Release-Note: >Audit-Trail: >Unformatted: To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200006232013.XAA67061>