Date: Wed, 03 Jan 2007 03:16:55 +0100 From: Torfinn Ingolfsen <torfinn.ingolfsen@broadpark.no> To: freebsd-multimedia@freebsd.org Subject: Re: MythTV testing results Message-ID: <20070103031655.5f49f816.torfinn.ingolfsen@broadpark.no> In-Reply-To: <c39ec84c0701021559i5923751fs5fbc3f1bcb71340d@mail.gmail.com> References: <459A4E82.6090201@xs4all.nl> <c39ec84c0701021559i5923751fs5fbc3f1bcb71340d@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 03 Jan 2007 00:59:36 +0100
usleepless@gmail.com wrote:
> this is the same problem as Torfinn reported. i think it has to do
> with the way ipc is done in mythtv ( ie: the problem is usleep() is
> not guaranteed to do a context-switch ).
I tried the quick'n' dirty hack you suggested, but it failed. (Report is
in the archives of this mailing list).
So far, I have not been able to find a way to fix it som MythTV
compiles with that hack (im not a developer, just a fix-it guy).
If you have any hints on how to proceed, I would like that very much,
if you can spare the time.
I must try to read up on Linux usleep and QT threading, to see if I can
figure this thing out.
> - this port ( mythtv-0.20 ) is patched quick 'n' dirty by redefining
> usleep() and a handfull(maybe more) of other patches ( i can be of
> help, and would love to do so )
> - i create diffs for my personal branch against 0.18-fixes and we!
> try to apply them to 0.20 ( either the ipc-patches only, or the
> ipc-patches and the pgsql patches ). this would be ideal.
My personal choice would be going for the "create patches to implement
ipc right", and then hang around on the mythtv develooer list and start
asking questions and poking them with the patches.
As this (ipc pathes) would be rather major, I don't think that the
mythtv developers would accept them at first, but perhaps over time we
could find other developers who are interested in MythTV on FreeBSD and
so would help to get the patches in. MythTV bills itself as "a project
by developers for developers", which puts some limits on thing like
user friendlyness in areas like: getting it to work, maintenance
(upgrading) and so on.
Anyway, if we manage to create ipc patches for MythTV, my guess is that
we would generate enough knowledge to be able to maintain them as
MythTV evolves, at least for some time.
Perhaps we could have a Wiki page somewhere with things for MythTV on
FreeBSD?
Starting out with the unanswered questions, we could "fill in the
blanks" as we learn about each new thing.
Kinda like this:
Q: Why doesn't MythTV 0.xx work on FreeBSD after I have patched it and
compiled it?
A: Because MythTV uses (Linux) usleep which as a side effect also does
context switching. In FreeBSD, usleep doesn't have this side
effect. To fix this, you need to reimplement ipc like this...
You get the idea.
--
Regards,
Torfinn Ingolfsen,
Norway
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070103031655.5f49f816.torfinn.ingolfsen>