glibc errors on SheevaPlug / Debian Squeeze

[bwrrp]

As previously mentioned in issue #2131, I'm having trouble getting node to work on my SheevaPlug with Debian squeeze. I'm working from the latest v0.8 branch checkout, and compiling node using

./configure --without-snapshot
 make CFLAGS+=-march=armv5t CXXFLAGS+=-march=armv5t

After this, running the tests results in several glibc errors:

*** glibc detected *** out/Release/node: free(): invalid next size (fast): 0x006a4040 ***
*** glibc detected *** out/Release/node: free(): invalid next size (normal): 0x007e3180 ***
*** glibc detected *** out/Release/node: free(): invalid pointer: 0x007ec678 ***

As most of these were in tls/https-related tests I tried passing --shared-openssl to configure. This reduced the number of glibc errors in the tests, but I'm still getting the error in test-os.js. Running this through gdb is unfortunately not very useful, as I seem to be hitting some bug where the stack trace is:

#1  0x403bdca4 in abort () from /lib/libc.so.6
No symbol table info available.
#2  0x403bdca4 in abort () from /lib/libc.so.6
No symbol table info available.
#3  0x403bdca4 in abort () from /lib/libc.so.6
No symbol table info available.

followed by the same abort() all the way down. Any suggestions are very welcome.

[bnoordhuis]

I have no idea what might be causing that. Does it make a difference when you set ulimit -c unlimited and inspect the core dump instead of a live binary?

[pedrolamas]

Seeing the same message with v0.8.5 on Arch Linux, running in a Raspberry Pi...

[docfaraday]

Pulled down the stable 0.8.6 tarball, and got this error. Ran the test under valgrind, and it seems that we're walking off the end of an array of size 0 down inside of uv_cpu_info() because /proc/cpuinfo is not formatted as expected. Fixing the code (just a sanity check in the loop on numcpus, with a decrement) prevents the crash, but the test still fails (I'm guessing because 0 cpus is something the test does not like).

[bnoordhuis]

@docfaraday Can you post the output of cat /proc/cpuinfo?

[docfaraday]

FWIW, here's my cpuinfo:

Processor       : ARMv7 Processor rev 2 (v7l)
processor       : 0
BogoMIPS        : 596.46

processor       : 1
BogoMIPS        : 599.22

Features        : swp half thumb fastmult vfp edsp thumbee neon vfpv3 tls 
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x1
CPU part        : 0xc09
CPU revision    : 2

Hardware        : OMAP4 Panda board
Revision        : 0020
Serial          : 0000000000000000

[bnoordhuis]

@docfaraday Thanks. Continues in joyent/libuv#526.

[bnoordhuis]

This is fixed in joyent/libuv@9a48381. It will be in the next v0.8 release.