Messages in this thread | | | From | Jared Mauch <> | Subject | Re: Problems raising fd limits in 1.3.78-1.3.80 | Date | Thu, 4 Apr 1996 11:50:42 -0500 (EST) |
| |
A lot of this stuff can (and [IMHO] I think) should be configurable via /proc/sys/* as some variable already are. The libc weenies could have their programs look at the file in /proc to find out the limits.
- Jared
Roman Gollent spewed: > > Stephen Tweedie wrote: > > Yup. I wonder if it's worth upping the limit in libc (as exported by > > linux/fs.h) to 1024 anyway? It only really affects the size of the > > fd_set, so it doesn't cost much. It ought to be pretty easy to make > > the kernel's maximum fd number configurable, with processes inheriting > > the current default value on fork() (we probably don't want to change > > the fd table size dynamically on existing processes). > > This brings up another point. How much of a rewrite is required to be > able to dynamically reconfigure certain kernel variables without > recompilation? > > I'm thinking of an /etc/sysconfigtab (Digital Unix) equivalent where > one can edit the parameters, reboot the system and have them take > effect. > > Roman >
| |