Date: Mon, 23 Feb 2004 13:42:14 -0700 From: Lance Gilbert <Lance@thefrontnetworks.net> To: freebsd-emulation@freebsd.org Subject: Implimenting New Linux System Calls Message-ID: <200402231342.14187.Lance@thefrontnetworks.net> In-Reply-To: <200402231901.i1NJ1aIL035110@freefall.freebsd.org> References: <200402231901.i1NJ1aIL035110@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi there, I'm new to the list, so excuse me if I bring up some topics which have been previously covered. (I just briefly looked through the mail archive) I have been working fairly extensively with the Linux emulation in FreeBSD over the past few months, primarily to get WineX 3.x working correctly under it. What concerns me, is that according to the response of 'uname -a' under Linux emulation, (chrooted) the implimentation is only at Linux kernel 2.4.2. With Linux already at 2.6.x and much later versions of 2.4.x, it would seem that FreeBSD's Linux emualation has started to fall behind in terms of syscalls. For example, I recently built a linux_base-Fedora port for myself. It uses a newer version of glibc then pervious linux_base. (glibc-2.3.2-101) As a result, even some simple commands like 'ls' core dump as a result of unimplemented system calls from newer linux kernels. I have tracked down the missing system call in question here, and it is an issue in "__pthread_initialize_minimal" from libpthread. (libpthread is provided by glibc) Point and case being, if things like 'ls' are core dumping now due to a lack of updated system calls, how long until this trend renders Linux Emulation useless, and what effect will the movement to 2.6 have here? Not to be completely glom-and-doom, I have started investigating the design of the Linux Emulation kernel module in order to start implimenting these calls myself, in an aim to keep it continually current with the current released Linux kernel. Is there already a team assigned to this task, or if not, would it perhaps be wise to start a project aimed at maintaining this? If anyone else is interested, I am also looking at the current state of linprocfs, and the challenges involved with updating that, especially given the fact that in 2.6 proc has changed fairly substantially. (Has there been any thought about how to manage 2.6 support while not breaking 2.4 support at this point?) My apologies for the long winded mail. Please feel free to respond to this on, or off list to me. -- Thanks, Lance Gilbert IT Department Head The Front Networks <Lance@thefrontnetworks.net>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200402231342.14187.Lance>