Re: Bug#896827: perl: FTBFS on riscv64: t/re/fold_grind timeout
2018-04-26 0:25 GMT+02:00 Manuel A. Fernandez Montecelo
<manuel.montezelo@gmail.com>:
> 2018-04-24 23:10 GMT+02:00 Manuel A. Fernandez Montecelo
> <manuel.montezelo@gmail.com>:
>>
>> I'm just firing up a build in the board to see if it passes the tests,
>> I will report when it finishes, if everything goes all right.
>
> So it didn't go all right for various reasons, trying again tonight
> but I am not very hopeful. For example I am using tmpfs, and some
> tests fail (or at least complain) due to that.
In the end it built successfully last night, things like the one about
tmpfs were only complaints.
All tests successful.
Elapsed: 3970 sec
u=57.60 s=9.86 cu=3460.11 cs=181.47 scripts=2445 tests=1244110
Time for the whole build:
Build needed 04:28:34, 541080k disk space
(This was in the board.)
> However, I think that the best thing to do is indeed increasing the
> time to perform the tests, to see if it advances or gets stuck in some
> tests, so please enable the change in the next upload.
>
> If it's not a problem for other architectures I'd even use a bigger
> factor, like 5 instead of 2. Some workloads under qemu are really
> slow, like 50x times slower than arm64 and 3-4x slower than m68k or
> sh4.
So the above is in the board, this still applies for the buildds.
Re-reading my paragraphs above I shall say, to not give a terrible
impression, that some workloads are not slower than in other
architectures.
But since longer builds are not a problem for us, as long as it
doesn't affect other architectures, it'll be better to increase the
factor for this arch, to be on the safe side.
Cheers and thanks.
--
Manuel A. Fernandez Montecelo <manuel.montezelo@gmail.com>
Reply to: