[Bug go/59431] [4.9 regression] runtime FAILs on Solaris

[ro at CeBiTec dot Uni-Bielefeld.DE]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59431

--- Comment #4 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #3 from Ian Lance Taylor <ian at airs dot com> ---
> Yeah, I didn't think that would fix this problem, I was just hoping for more
> consistent error messages--e.g., "out of memory" rather than "caught signal
> while mallocing: 10".

I've now looked a bit closer and found that runtime_SysAlloc is indeed
failing.  The problem seems to be documented behavior of mmap(2) in
32-bit processes on Solaris:

     The size requested is incremented by the 16 Kbytes for these
     pages  and is then subject to rounding constraints. The con-
     straints are:

         o    For 32-bit processes:

                If length > 4 Mbytes
                        round to 4-Mbyte multiple
                elseif length > 512 Kbytes
                        round to 512-Kbyte multiple
                else
                        round to 64-Kbyte multiple

E.g. for os/check, I find almost 50000 512-byte allocations, each of
which costs 64 kB of virtual address space.  This will quickly exhaust 4
GB, while not being a noticable problem for 64-bit processes.

    Rainer