Re: HEADS-UP: starting to commit linuxolator (SoC 2006) changes...

From: Divacky Roman <xdivac02_at_stud.fit.vutbr.cz>
Date: Tue, 15 Aug 2006 22:48:17 +0200
On Tue, Aug 15, 2006 at 01:04:52PM -0700, Kip Macy wrote:
> What does testing buy us when you've already admitted that futexes et
> al. are broken?

the futexes are not 100% working which is different from broken.

it passes my pthread tests and its able to run quite complicated programs (firefox, opera)

the problem is (for example) with realplay which hangs waiting on a futex when I attempt
to play a file. I noticed that this behaviour changes based upon what pid I copy out
to the parent thread (when I copy linux-tid ff/opera works and realplay not, when I copy
linux-pid ff is broken, realplay works). if anyone has an idea what can cause this
feel free to tell me.

the point is - other people might see what the bug is. I dont see it.

what commiting this code into src means?

1) more eyeballs to review 

2) coverity run - I strongly suggest extending coverity to run on some p4 branches

3) some panics etc. might be revealed

4) the commit is almost 100% self-contained and doesnt break anything

while I agree that Alexander could wait 2-3 days in general case I also know his reasons
why he commited it so fast - his vacation ends and he wanted to devote today to integrating
this stuff into the cvs. On the other hand I've seen too many patches which got never integrated
that I am glad this one is different.

my 2 cents

Roman Divacky (the author of the patch in question)
Received on Tue Aug 15 2006 - 18:48:25 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:59 UTC