Re: [Valgrind-users] Failing to start valgrind

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

> % cat /proc/self/maps
> 08048000-0804a000 r-xp 00000000 08:02 210862     /bin/cat
> 0804a000-0804b000 rw-p 00002000 08:02 210862     /bin/cat
> 0804b000-0804c000 rwxp 00000000 00:00 0
> 15556000-1556a000 r-xp 00000000 08:02 323984     /lib/ld-2.3.2.so
> 1556a000-1556b000 rw-p 00014000 08:02 323984     /lib/ld-2.3.2.so
> 1556b000-1569c000 r-xp 00000000 08:02 323939     /lib/libc-2.3.2.so
> 1569c000-156a1000 rw-p 00130000 08:02 323939     /lib/libc-2.3.2.so
> 156a1000-156a4000 rw-p 00000000 00:00 0
> 3fffe000-40000000 rwxp fffff000 00:00 0
> % 

> MontaVista(R) Linux(R) Carrier Grade Edition 3.1
>> Linux/i686 2.4.20_mvlcge31_p2-langley5

It seems that the kernel configuration parameter TASK_SIZE in this
Linux kernel is 0x40000000 (1GB), because the stack ends there:
    3fffe000-40000000 rwxp fffff000 00:00 0
This means that the user virtual address space is restricted
to 1GB (0 to 0x40000000) per process.  Probably this kernel is
targeted to supporting a very large number of medium- to small-sized
simultaneous processes, using a large amount of physical RAM
(many GB).  The kernel and each process share address space;
the more physical RAM, the more the kernel needs address space
to keep track of it, and to supervise many simultaneous I/O streams.

Valgrind assumes that the user virtual address space is at least
11/4 GB (0xb0000000), plus a little more.  The kernel TASK_SIZE is
typically 0xc0000000 or 0xf0000000.  This is common in many
desktop and small server configurations with at most a
few GB of physical RAM.

-- 