Roland McGrath <roland(a)hack.frob.com> writes:
> Another way to fix this is to simply remove the whole block
> special-casing singleton item. The general code handles the case of
> nitems == 1 just fine, it looks like that block is simply an
> optimization.
It is not simply an optimization, it's calling handle_core_item
differently. The 2008-08-01 log entry shows I added that special case
Right, the '\n' format has to be called with a repeated size. I missed
that.
specifically for the case where handle_core_item gets the non-null
last
argument and handles repeats itself. I really don't recall the details
any more, but what I can surmise is that the one case that would use
this is "VMCOREINFO" notes. Those appear in core files generated by
makedumpfile from /proc/vmcore data in the kdump stuff. Can you try to
generate such a file that we can use to see what I was going for? Show
how eu-readelf -n output for such a file is affected by any change you
propose.
My fix did not impact the formatting:
VMCOREINFO 1366 <unknown>: 0
OSRELEASE=2.6.35.11-83.fc14.x86_64
PAGESIZE=4096
SYMBOL(init_uts_ns)=ffffffff81a4c5b0
SYMBOL(node_online_map)=ffffffff81b840b0
[... etc ...]
There's about 50 lines total.
So, the vmcore that I have in my hands presumably contains a lot of GPL
code, and besides it has close to 2 GB. But how about snipping just the
ELF header and core note data, and converting it to a test case? Would
that be acceptable? Do you think it's acceptable from licensing
perspective?
Thanks,
PM