[ccan] block_pool tests fails on OpenIndiana (with suncc)

[Rusty Russell]

Hi Joey,

        Random testing obviously isn't sufficient.  Looks like you
didn't take into account zero-length blocks properly?

I changed the NULL to stdout, and here's the results:

1..1
Allocating 10000 blocks...
Finished allocating
    10000 blocks
    6049581 bytes
    778 pages
Making sure block ranges are unique...
Allocations 107 and 108 overlap

gdb gives:

Breakpoint 2, test_block_pool (blocks_to_try=10000, out=0x8073090, verify_heap=1)
    at /home/amitay/devel/ccan/ccan/block_pool/test/run.c:124
124                     print("Allocations %zu and %zu overlap\n", i-1, i);
(gdb) p *a
$1 = {size = 20, ptr = 0x808de24 'U' <repeats 200 times>...}
(gdb) p *b
$2 = {size = 0, ptr = 0x808de24 'U' <repeats 200 times>...}
.

-- 
  How could I marry someone with more hair than me?  http://baldalex.org