Код: Выделить всё
http://lists.freebsd.org/pipermail/freebsd-bugs/2008-July/031568.html
http://www.freebsd.org/cgi/query-pr.cgi?pr=125563
))))= ))) ггггюзайте с HEAD'a re для RELENG_7 и будет вам счастье
чего написал, сам то понял?юзайте с HEAD'a re для RELENG_7 и будет вам счастье
Кроме написания PR, попробуйте связаться с Hans Petter Selasky <hselasky@c2i.net>. А то PR может лет 5 висеть и никто к нему не притронется...paradox писал(а): пойду HID реппорт офформлять
как токо лис протестит окончательно
хотя и так ясно что поможеться
)) это разработчик написал мне> no. I about reason for split driver to 2 device
> driver written by Realtek - one driver for all device...
Ahh, yes that's good question.
rl(4) supports just dumb plain controllers and the hardware has
a lot of limitation such as lack of DMA(!!!), any kind of hardware
acceleration, too small FIFO and small number of descriptors.
Whereas re(4) supports more advanced hardware created by RealTek.
These controllers have some kind of hardware assistance(You may
already know they are also buggy).
The fundamental problem with merging the two driver into one comes
from complete different register usage/layouts and descriptor
structures. RealTek is notorious for its continuously changing its
internal register layouts/descriptors for each revision. Of course
it could be handled in driver with lots of switch or if statements
as vendor's FreeBSD driver did but that will significantly reduce
readability of code. With the advent of the second generation of
PCIe based controllers from ReakTek, re(4) also suffers from these
different kind of registers and descriptor handling for newer
controllers.
If you had chance to see vendor's driver you wouldn't say the
possibility of single driver that is able to serve all kind of
RealTek controllers. If you had checked Linux driver released by
vendor they also release dedicated drivers for each family(8169,
8168, 810x, 813x etc).
P.S. RealTek didn't write a driver for FreeBSD. They just copied
our rl(4) driver and added some access functions to their
undocumented registers to recognize recent controllers.
vendor's driver makes your hardware just like a plain 8139
controller.
