On Sunday 18 September 2005 22:46, Leonard Orb wrote:
> klickmich wrote:
> > Probably a mismatch in the driver tools versions. Did you make sure to
> > always use the tools that come with the respective driver?
>
> Yes, I did. But it should work regardless of the tool's version, there are
> only ioctls to be done, and as long as the parameters for the ioctl do not
> change it makes no difference what program uses it.
That's no true for this particular utility: it uses several internal ioctls
that can and do change. These ioctls can be found in the ivtv.h header under
#ifdef IVTV_INTERNAL. The non-internal ioctls do not change (in principle :-)
of course.
> Back to the problem:
>
> I just wanted to start testing 0.3.1 and 0.3.1z, but both versions won't
> compile: with 0.3.1 gcc spits warnings about 'pci_dma_sync_single' which
> appears not to exist in the kernel 2.6.11.12 (or wherever it is supposed
> to be). Later during install the warning is '*** Warning:
> "pci_dma_sync_single" [/usr/src/ivtv-0.3.1/driver/ivtv.ko] undefined!'.
It's a kernel change, here's a snippet from ivtv-queue.c that shows the
workaround. Basically pci_dma_sync_single has to be replaced with
pci_dma_sync_single_for_device. A #define might do the trick here to get it
compiling quickly.
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 8)
pci_dma_sync_single_for_device((struct pci_dev *)itv->dev,
dma->SG_handle,
(sizeof(struct ivtv_SG_element) *
IVTV_DMA_SG_OSD_ENT),
PCI_DMA_TODEVICE);
#else
pci_dma_sync_single(itv->dev, dma->SG_handle,
(sizeof(struct ivtv_SG_element) *
IVTV_DMA_SG_OSD_ENT), PCI_DMA_TODEVICE);
#endif /* LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 8) */
>
> 0.3.1z has a compile time error, I did not check for a quick work around
> as i did some times before with other probs because saa7115 is an active
> module for the pvr 250. Btw version 0.3.0l has the same problem:
> --->
> In file included from /usr/src/ivtv-0.3.1z/driver/saa7115.c:83:
> /usr/src/ivtv-0.3.1z/driver/ivtv.h:57: field `data' has incomplete type
> <---
struct ivtv_sliced_data {
unsigned long id;
unsigned long line;
unsigned char data[];
};
'data[]' is a gcc extension. Which gcc version are you using? I seem to
remember reading something somewhere that this is no longer supported in the
new 4.0 gcc compiler or somthing like that. Try replacing data[] with
data[1], or see if the compiler has a flag allowing this construct.
>
> pci_dma? Could it be that some changes in dma handling were necessary
> during 0.3.0 or 0.3.1 and that the problem was intruduced then?
>
> If the drop outs are caused by a too slow dma irq handler or such this
> could explain that the problem already existed in 0.3.3z and became worse
> in 0.3.5z?
I've got no idea where to look. First I need data such as you are providing
(and many, many thanks for that). After 0.3.1z you can try 0.3.1 (the 0.3.0
series was a dead end, no point checking that). 0.3.1 was actually split off
from 0.2.0-rc2d. I hope it compiles...
If it takes too much time to get these old versions to work then don't bother.
I'm more interested in the precise point between 0.3.4z and 0.3.5z where the
problem got worse. That might give us an idea where to look to solve the
remaining problem. Also, the minor problems that you see with the older
version may have been fixed in the latest version, except you don't see it
because of the more severe bug that was introduced.
Hans
Hans
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
ivtv-devel mailing list
ivtv-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ivtv-devel
