On Tue, May 11, 2010 at 08:23:41PM +0100, Robert Watson wrote: > > On Mon, 10 May 2010, Lev Serebryakov wrote: > > > I'm proting some application from Linux, which discover its stack bounds > > by reading and pasing "/proc/self/maps". FreeBSD have > >"/prov/curproc/map", but I can not find how to determine which record is > >for stack (I've looked into implementation of proc_fs, but it doesn't > >contain any specail processing for process stack). > > > > How could I determine stack bounds of current process on FreeBSD 7/8/9? > > The "procstat -v" command in 8.x and 9.x will give this information based > on sysctls; we're about to integrate a libprocstat(3) library which will > provide a public API for this information. I'd agree with Kostik that you > should think carefully about whether the application really needs this > information :-).
Unfortunately, it is not that simple. How to guess which vm_map_entries are from the stack ? To complicate the issue, the stack is usually fragmented, i.e. continuous VA area is covered by several adjanced entries. The answer "look at the kern.ps_strings" is bad as well, since it gives wrong answer e.g. for ia32 binary on amd64. Idea to look at the highest mapped address and then descend might be safest, but as I already pointed out, libthr.so clamps the main stack to keep its size the same as for non-main threads. And this is ignoring issues of non-main thread stacks, as well as signal altstacks. As I said, there is no good answer to the question, and better strategy is to understand why application need this.
pgpZoJGshRGhq.pgp
Description: PGP signature
