Last version ( STAPI version ) cardreader not work.

[ngc1927]

Revision

r8032

Issue Description

Not work card reader ( internal )

When the issue occurs

Always with last commit ( r8025 work ok )

[ni hao]

attach complete (!) d255 log. Here card reader works, so discuss first in forum

[ngc1927]

Now attached 255 debug

@theparasol
I test last 8039 ..not work.

[Admin]

Possibly cause by r8028?

[ngc1927]

Now have tested r8027 ( work ok ) and r8029 ( not work ).

P.S.
have jump r8028 why not compile with SH4

[theparasol]

Try http://www.streamboard.tv/oscam/changeset/8043

Actually reverted the whole read all chars at once for sh4 only.
If I got it right your box will work again.
Perhaps a more experienced dev can rewrite the code so that it pulls all chars from reader of SH4. But perhaps these kind of boxes only deliver chars one by one?

[Admin]

What is somehow strange from the log is that the error exists when he just requests a chunk of size 1?

[theparasol]

Yup, saw that too and couldnt find a solution, therefor the revert for SH4/STAPI :(

[ngc1927]

Also r8043 not work, with other error ( see log attached 8043.. )

[theparasol]

Double checked it again... only thing I found was a readed = 0
I missed to reimplement. If this not fixes it I'm out of options and somebody else better give a look at it.

http://www.streamboard.tv/oscam/changeset/8049

[Admin]

I know the problem in the original commit I think:

while( (((tv_spent.tv_sec-tv.tv_sec)*1000000) + ((tv_spent.tv_usec-tv.tv_usec)/1000000L)) < (time_t)(timeout))

{

readed =read(reader->handle, &data[count], size-count);
gettimeofday(&tv_spent,0);

}

The break condition and the "count +=readed;" is missing here! That's why after the first successful receive, it tries to read further until timeout time is reached and at this point "read" probably returns 0 then resulting in the error!

[theparasol]

If you can fix it, go ahead. BTW: I like it the way its now: splitted up in section SH4 and other devices. Much better readable code

[Admin]

Thought on something like that.

[kroki]

You will switch the codeblocks on the _sh4_ Pragma, this affects all Sh4-Versions.Not only the Stapi-Readers!
Not all Sh4-Versions are using the Stapi-Interface.

[ngc1927]

r8052 work ok ..thx

[Admin]

Please test the attached patch first. If it works, it will yield better performance.
@kroki: Well, but you can compile without stapi even if you are on such a box...you mean we should check for internal reader additionally?

[kroki]

@Admin: That's right. If you use a Enigma2 or Neutrino-Image on these Boxes you don't use the Stapi-Interface. So with your Patch you always mention, that sh4 = stapi.
So the optimisations for the other readers didn't have any effect on sh4, if you use the pragma _sh4_. Better use WITH_STAPI.

[ngc1927]

@Admin
I applied your patch to r8048 ..and work ok

[theparasol]

Great, thats the best solution of all.
Tnx for testing!

[Admin]

I'll try to make it more restrictive in the evening...please don't commit yet.

[Admin]

So here's the revised patch. The code path is only valid for WITH_STAPI and only for the internal readers => other serial readers on these plattforms should work with the regular code.

[Admin]

Please give feedback so that I may commit it.

[kroki]

For me, without stapi it works ....

[Admin]

Committed now in r8069.

[ngc1927]

Sorry ..i see too late your request of test, ..yes in stapi work ok.
Thx @theparasol and @Admin

[N3m3515]

after r8069 my sh4 without stapi lags on reading on my internal reader. slow initialisation and reader timeouts with the internal reader because my times are over 5000ms. on 8068 everything works fine. everything is reported in this thread: http://www.streamboard.tv/wbb2/thread.php?threadid=36758&page=3

[N3m3515]

alborland found a fix:

http://www.streamboard.tv/wbb2/thread.php?postid=446190#post446190

can someone include this?

[Admin]

Fixed in r8113.