Per the advice of Jens Axboe, I have added one line to my
previous fix to make linux-2.5.1-pre7/drivers/block/xd.c compile
(the "&& (CURRENT->flags & REQ_CMD)" line). Here is the diff
against pristine linux-2.5.1-pre7/drivers/block/xd.c.
There is no maintainer for xd.c listed in linux/MAINTAINERS,
although I am and have been mailing to the email addresses that appear
in xd.c.
Linus, if nobody says otherwise, I recommend that you apply
this patch.
--
Adam J. Richter __ ______________ 4880 Stevens Creek Blvd, Suite 104
[email protected] \ / San Jose, California 95129-1034
+1 408 261-6630 | g g d r a s i l United States of America
fax +1 408 261-6631 "Free Software For The Rest Of Us."
On Sat, 8 Dec 2001, Adam J. Richter wrote:
>
> Linus, if nobody says otherwise, I recommend that you apply
> this patch.
Well, I already applied your previous one, in fact, and it's in the
just-uploaded pre8 kernel. Mind verifying that and sending the incremental
update?
Btw, do you actually _have_ a machine that uses the xd driver, or was this
patch done just out of some perverse joy in theoretical retrocomputing?
Linus
>> Linus, if nobody says otherwise, I recommend that you apply
>> this patch.
>Well, I already applied your previous one, in fact, and it's in the
>just-uploaded pre8 kernel.
Thanks!
>Mind verifying that and sending the incremental update?
No problem. I have attached the one line addition below.
>Btw, do you actually _have_ a machine that uses the xd driver, or was this
>patch done just out of some perverse joy in theoretical retrocomputing?
Years ago, I submitted a patch to allow configuration of
the kernel with "./configure", which would configure every driver as
a module, aside from compiling in the initial ramdisk and the initial
ramdisk's filesystem. I haven't configured a kernel for years; the
boot scripts and hot plugging software do that. The same binary
build of the kernel can run on all of my x86 hardware.
That is why I know that 92 files failed to compile on x86 in
2.5.1-pre7, and largely why I care about fixing xd.c.
Anyhow, thanks for asking. By the way, if you have any interest
in integrating my "./configure" functionality now, I would be happy to
clean it up and resubmit it. (It is mostly a patch to scripts/Configure.)
Adam J. Richter __ ______________ 4880 Stevens Creek Blvd, Suite 104
[email protected] \ / San Jose, California 95129-1034
+1 408 261-6630 | g g d r a s i l United States of America
fax +1 408 261-6631 "Free Software For The Rest Of Us."
--- linux-2.5.1-pre8/drivers/block/xd.c Sat Dec 8 21:29:50 2001
+++ linux/drivers/block/xd.c Sat Dec 8 20:19:54 2001
@@ -287,6 +287,7 @@
INIT_REQUEST; /* do some checking on the request structure */
if (CURRENT_DEV < xd_drives
+ && (CURRENT->flags & REQ_CMD)
&& CURRENT->sector + CURRENT->nr_sectors
<= xd_struct[MINOR(CURRENT->rq_dev)].nr_sects) {
block = CURRENT->sector;
On Sat, 8 Dec 2001 22:06:57 -0800,
"Adam J. Richter" <[email protected]> wrote:
> Anyhow, thanks for asking. By the way, if you have any interest
>in integrating my "./configure" functionality now, I would be happy to
>clean it up and resubmit it. (It is mostly a patch to scripts/Configure.)
Already in kbuild 2.5, it supports make allyes, allno, allmod, random.
On Sat, 8 Dec 2001, Linus Torvalds wrote:
> On Sat, 8 Dec 2001, Adam J. Richter wrote:
> >
> > Linus, if nobody says otherwise, I recommend that you apply
> > this patch.
>
> Btw, do you actually _have_ a machine that uses the xd driver, or was this
> patch done just out of some perverse joy in theoretical retrocomputing?
I have somewhere a machine that used to work with such hardware, but
have never tested it with anything newer than 2.2.x kernel and have no
time to perform such tests in the near future (Maybe I'll find some after
Christmas).
AFAIR the hardware works fine with 386/486 with clock up to 66 MHz.
Faster machines have problems with BIOS initialization, probably due to
very slow EPROM chips or badly designed timing calculations in their
BIOSes. They *might* work with BIOS disabled/romoved, but all hardware I
have has the BIOS chips integrated.
Most hardwate supports drives up to 40 MB (I have only 20s) and the
transfer rates about 20-40 kB/s. Faster (with memory mapped I/O) boards
are not supported by the driver.
--
=======================================================================
Andrzej M. Krzysztofowicz [email protected]
tel. (0-58) 347 14 61
Wydz.Fizyki Technicznej i Matematyki Stosowanej Politechniki Gdanskiej