Alex Woods wrote:
The case is, if I do
ifconfig dvb0_0 promisc ,
the kernel panics crying about a bug in sched.c (!).
That looks like the same bug i have in the ttusb-nova driver, that the
section handler is directly called from the (interrupt-) urb-complete
handler. This has to be fixed in the
Hello Gregoire,
I just updated the CVS and try to compil against 2.5.70:
Thanks again for your tests -- you are very quick. 8-)
LD drivers/media/common/built-in.o
CC drivers/media/dvb/dvb-core/dvbdev.o
In file included from drivers/media/dvb/dvb-core/dvbdev.c:35:
Hi,
please use tasklets to do the real work for your interrupt handlers.
Take a look in the PCI drivers for an example, the pretty simple to
convert: just schedule your tasklet with old irq handler function in
the new irq handler.
Since tasklets are strongly serialized (if the tasklet is
Felix Domke wrote:
Hi,
please use tasklets to do the real work for your interrupt handlers.
Take a look in the PCI drivers for an example, the pretty simple to
convert: just schedule your tasklet with old irq handler function in
the new irq handler.
Since tasklets are strongly serialized (if
Felix Domke wrote:
I for myself would like to have the lower dvb driver layer being able to
handle calls from interrupt context. tasklets always involve a higher
latency, and sometimes (like in our case) it's difficult to buffer the
data away. Or is there a strong reason not to support this?
On Wed, May 28, 2003 at 10:26:34AM +0200, Michael Hunold wrote:
The devfs stuff has been simplified in the last kernel versions,
devfs_handle_t is now gone. This causes the compilation to fail and
has to be worked around.
It is not trivial to work around this in dvb_compat. With a heavy