On 27 November 2010 19:29, Brian Harring <ferri...@gmail.com> wrote: > Look through linux-user/syscall.c; looks like the flaw is more in do_brk > itself. Invocations of do_brk *appear* to all assume that it's > basically brk like in it's behaviour- -1 on failure, else a non-negative > value of what the size now is.
I think this is wrong. The primary user of do_brk() is the Linux user system call emulation, so do_brk()'s semantics follow those of the kernel implemented syscall. (This is different from those of the brk() library routine, hence the confusion.) That means it returns the old brk value on failure, not -1. [This is documented in the NOTES section of the Linux brk(2) manpage, or you can look at the kernel sources.] You could argue that this is a bit unhelpful for platform-independent code like the ARM semihosting routines which end up coding to an API which might be platform-dependent between linux-user and some-other-user modes... -- PMM
