[lld] r319518 - Add an additional test for r319503.

[Rafael Avila de Espindola via llvm-commits]

Thanks!

Rui Ueyama <ruiu at google.com> writes:

> I don't have time at the moment too, so filed as
> https://bugs.llvm.org/show_bug.cgi?id=35536 to not forget this.
>
> On Tue, Dec 5, 2017 at 8:38 AM, Rafael Avila de Espindola <
> rafael.espindola at gmail.com> wrote:
>
>> Rui Ueyama <ruiu at google.com> writes:
>>
>> > 960 and 1031 aren't that different, and not using the prime number
>> > shouldn't make change as long as the lengths of hash chains are not too
>> > biased (which I believe true).
>> >
>> > I took a quick look at `do_lookup_x` function in elf/dl-lookup.c in
>> glibc,
>> > and looks like the penalty of hash collision is pretty low there (we just
>> > need to skip a uint32_t value for each collision.)
>> >
>> > So the difference is interesting.
>>
>> I honestly don't have the time to test it further, but it is pretty
>> simple to experiment if you are interested:
>>
>> * Build a regular release of llvm+clang+lld in directory build.
>> * Mount a tmpfs in build2
>> * Use build to configure (cmake) a llvm in build2 with BUILD_SHARED_LIBS
>> * Run ninja check-llvm
>> * Run perf stat/record llvm-lit ... (copy and past the command from
>>   ninja -v check-llvm)
>>
>> Cheers,
>> Rafael
>>