Re: [B2-devel] problems caused(?) by recent patches

[Charles S. <bas...@ch...>]

Unfortunately, I can’t get it to reproduce on my MacBook Pro, no matter how I change the settings, so it’s a little hard to diagnose at this point. But if it’s crashing even on Intel machines, I guess the best thing to do would be to disable it for the time being until it is possible to figure out what is going on.



Charles

On Jan 14, 2010, at 8:00 PM, Alexei Svitkine wrote:

> Charles,
> 
> Can you investigate the crashes with your precise timer patch?
> 
> -Alexei
> 
> On Thu, Jan 14, 2010 at 8:58 PM, Alexei Svitkine
> <ale...@gm...> wrote:
>> I've reverted the SDL audio patch.
>> 
>> On Thu, Jan 14, 2010 at 3:06 PM, Ronald P. Regensburg
>> <ron...@xs...> wrote:
>>> In answer to the comments by Charles Srstka:
>>> 
>>> About the precise timer patch:
>>> The crash with the precise timer crash was not limited to PPC
>>> machines. I noticed it myself on a PPC machine first while it did not
>>> happen on my Intel machine. However, within 6 days after posting my
>>> 18-10-2009 build I got in the SheepShaver forum four reports of the
>>> crash, each at the very same moment in the startup process, one on a
>>> PPC machine and three on a Intel machine. The 25-10-2009 build
>>> (without the precise timer patch) solved the crash for myself on my
>>> PPC machine as well as for the other reported PPC machine and for two
>>> out of the three reported Intel machines. On the third reported Intel
>>> machine the 25-10-2009 crashed with a SIGSEGV. That crash was resolved
>>> by enabling "Ignore Illegal Memory Accesses" (ignoresegv true). Later,
>>> there were a few more reports of SIGSEGV crashes on Intel machines
>>> with the 25-10-2009 build that could be solved this way. So far, I got
>>> one report of the 25-10-2009 build crashing with a SIGSEGV in Leopard
>>> on a PPC machine, not yet solved nor understood.
>>> 
>>> About the sdl-audio patch:
>>> Probably best to simply remove the patch. In the original
>>> pre-19-02-2009 BasiliskII/src/SDL/audio_sdl.cpp file the value for
>>> audio_spec.samples is set to 4096, well within the recommended range.
>>> 
>>> Ronald.
>>> 
>>> 
>>> Op 14 jan 2010, om 17:11 heeft Charles Srstka het volgende geschreven:
>>> 
>>>> On Jan 14, 2010, at 7:18 AM, Ronald P. Regensburg wrote:
>>>> 
>>>>> 
>>>>> In Emaculation.com forums, Alexei Svitkine asked me to mail to this
>>>>> list about patches I needed to revert before building SheepShaver
>>>>> (and
>>>>> BasiliskII) for Mac OS X in order to avoid problems.
>>>>> 
>>>>> 
>>>>> The problems with the "precise timer patch" were discussed in the
>>>>> basilisk-devel list between October 23 and October 28. Those problems
>>>>> were the reason for me to make the 25-10-2009 build for
>>>>> Emaculation.com within days after the 18-10-2009 build.
>>>>> 
>>>>> In my last October 25 message to the basilisk-devel list I told which
>>>>> files I used for the 25-10-2009 build to omit the "precise timer
>>>>> patch" and the "sdl-audio patch".
>>>>> 
>>>>> The "sdl-audio patch" apparently made sound for most users worse
>>>>> rather than better. Users of both BasiliskII and SheepShaver builds
>>>>> created after that patch was added, complained about the problems for
>>>>> months and reported considerable improvement when I posted builds of
>>>>> both without the patch. That happened before I joined this mailing
>>>>> list. The problems were sound delays, stuttering, even skipping of
>>>>> sounds. The problem was more serious in BasiliskII than in
>>>>> SheepShaver.
>>>>> 
>>>>> I do not know whether these patches actually cause the problems or
>>>>> just make bugs in other parts of the code apparent.
>>>> 
>>>> The crash with the precise timer patch was mostly limited to running
>>>> the app on PPC machines if I remember correctly, wasn’t it? Since
>>>> SheepShaver runs completely different code paths on PPC and Intel,
>>>> with the PPC version running native code and the Intel using
>>>> emulation, I can see how the change could have caused problems with
>>>> one that would not have come up on the other. The best solution in
>>>> this case may be to enable the precise timer patch only when
>>>> building for Intel.
>>>> 
>>>> The SDL audio patch, on the other hand, is simply ill-advised in my
>>>> opinion. The docs for SDL recommend using a value between 512 and
>>>> 8192 for the ‘samples’ field of an SDL_AudioSpec. The patch in the
>>>> SheepShaver source sets it to 16384. The result is laggy audio. I
>>>> think the patch was made to reduce the demands on processing power
>>>> for older machines, but for those with modern hardware it seems to
>>>> cause more problems than it solves.
>>>> 
>>>> Charles
>>>> ------------------------------------------------------------------------------
>>>> Throughout its 18-year history, RSA Conference consistently attracts
>>>> the
>>>> world's best and brightest in the field, creating opportunities for
>>>> Conference
>>>> attendees to learn about information security's most important
>>>> issues through
>>>> interactions with peers, luminaries and emerging and established
>>>> companies.
>>>> http://p.sf.net/sfu/rsaconf-dev2dev
>>>> _______________________________________________
>>>> basilisk-devel mailing list
>>>> bas...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/basilisk-devel
>>>> 
>>> 
>>> 
>>> ------------------------------------------------------------------------------
>>> Throughout its 18-year history, RSA Conference consistently attracts the
>>> world's best and brightest in the field, creating opportunities for Conference
>>> attendees to learn about information security's most important issues through
>>> interactions with peers, luminaries and emerging and established companies.
>>> http://p.sf.net/sfu/rsaconf-dev2dev
>>> _______________________________________________
>>> basilisk-devel mailing list
>>> bas...@li...
>>> https://lists.sourceforge.net/lists/listinfo/basilisk-devel
>>> 
>> 
> 
> ------------------------------------------------------------------------------
> Throughout its 18-year history, RSA Conference consistently attracts the
> world's best and brightest in the field, creating opportunities for Conference
> attendees to learn about information security's most important issues through
> interactions with peers, luminaries and emerging and established companies.
> http://p.sf.net/sfu/rsaconf-dev2dev
> _______________________________________________
> basilisk-devel mailing list
> bas...@li...
> https://lists.sourceforge.net/lists/listinfo/basilisk-devel