This mail is an automated notification from the patch tracker
of the project: The GNU Hurd.
/**/
[patch #2507] Latest Modifications:
Changes by:
Daniel Wagner <[EMAIL PROTECTED]>
'Date:
Robert Millan <[EMAIL PROTECTED]> writes:
> At this point, I think it'd be better to use Oskit instead. To the people
> who have played with Oskit already, do you think it's viable to use it
> as a backend for userspace drivers?
Well, Oskit provides lot's of drivers but also forces you to emulate
Here is my second attempt. If this interface is somewhat sane I start
to hack the needed wrapper code.
thanks,
daniel
/*
* Copyright (c) 1997-1998,2002 University of Utah and the Flux Group.
* All rights reserved.
* @OSKIT-FLUX-GPLUS@
*/
/*
* Definition of a COM interface to support basic
Roland McGrath <[EMAIL PROTECTED]> writes:
> Nice work! Can you show the code and say something about the problems and
> hacks you had to do?
Sure (I was hopping to conceal my cruel hacks :) I modified the oskit
sio driver so that only ttyS1 is initialised and ttyS0 left alone so
that remote d
> for a low-level interface. My thinking is to come up with a reasonable
> oskit interface for serial ports with modem control (layered on top of the
> trivial streamdev, but different from ttydev).
I finally managed to get the sio driver from freebsd working with the
gnumach kernel. Was quite
The oskit_guid in your i8042 patch is already used by the socket
interface (oskit/net/socket.h).
#define OSKIT_STREAMDEV_IID OSKIT_GUID(0x4aa7dfad, 0x7c74, 0x11cf, \
0xb5, 0x00, 0x08, 0x00, 0x09, 0x53, 0xad, 0xc2)
#define OSKIT_SOCKET_IID OSKIT_GUID(0x4aa7dfad, 0x7c74, 0x11cf, \
> BTW looks this setup correct?
>
> hurd:~# showtrans /dev/ttyS1
> /hurd/term /dev/ttyS1 device @isa@0x02f8
I'm quite sure that "@isa@0x02f8" is wrong. The syntax for bus tree
walk is for me a bit cryptic (oskit/unsupported/bus_walk_lookup.c).
oskit_dump_devices() dumps following tree fo
> You need a lot more glue work in oskit-mach to support a set of control
> interfaces sufficient for serial devices.
gnumach/include/device/tty_status.h describes the interface which is
used for term. In the todo list is mentioned that this interface is
somewhat limited. My question is should
Marcus Brinkmann <[EMAIL PROTECTED]> writes:
> I think this usually points to a small typo in the GRUB boot script. Mach
> is extremely picky about it, and it doesn't provide useful diagnostics on
> certain errors. Fixing that would be nice.
A small workaround could be to provide a small script
> This is really interesting. What NIC's do you use?
Both (linux box and hurd box) have rtl8139 chips.
daniel
___
Bug-hurd mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-hurd
Marcus Brinkmann <[EMAIL PROTECTED]> writes:
> It's hard to believe you get no package loss in GNU Mach 1. Do you flood
> the GNU/Hurd box from a GNU/Linux box? I have seen horrible package loss in
> that situation, which was much better with GNUMach v2.
I see the same situation here. When I fl
Joachim Nilsson <[EMAIL PROTECTED]> writes:
> In comparison to what else is missing in the Hurd right now it is
> surely a bit unimportant. And I don't mean to belittle your work
> in any way!
No offense taken.
> I don't think your patch is wrong - it looks quite good actually.
> It might just b
Joachim Nilsson <[EMAIL PROTECTED]> writes:
> When I flood ping GNUmach2 I get 1% loss and no package loss with
> GNUmach1. Maybe we can spur some interest in tuning these numbers
> a bit?
Hmm, I don't know if this very important. The TCP/IP protocol handles
very well any packet loss :) I rath
ve decided to fix this before going into the serial stuff
again.
[1]: http://mail.gnu.org/archive/html/bug-hurd/2002-11/msg00187.html
2003-01-28 Daniel Wagner <[EMAIL PROTECTED]>
* i386/i386/spl.S (SOFTINT): New marco.
(spl0): Removed old Linux soft interrupt invokation
Hi
I'm try to get the serial driver from OSKit working with gnumach2.
Basically I just changed the Makefile to include the freebsd driver
library and added the oskit_freebsd_init_sio call to
gnumach/oskit/ds_osenv.c. Of course things are not that simple and so
I see a panic as soon the ttyS1 devi
Joachim Nilsson <[EMAIL PROTECTED]> writes:
> 1. Change the OSKit version dependency in configure.in to,
> at least, NEEDED_OSKIT_VERION=20010214. The changes for
Yes, that's not a bad idea.
> 2. When I build an optimized kernel, only IDE and one
> ethernet card, e
Hi
Marcus has asking me to post the patches which are needed to get a
'stable' kernel using pcmcia. These patches disable some code which
trigger some fatal bugs and therefor they are just quick and dirty
hack. They don't solve the problem.
I have uploaded a new version of the pcmcia patch. Now t
I've spend some time to investigate the panic I reported
earlier. Here is another bt, one I saw very often.
#0 panic (fmt=0x2328b6 "\r\npanic in device driver!")
at ../../../oskit-20020317+pcmcia/libc/gen/panic.c:31
#1 0x128551 in log_vpanic () at ../../gnumach/oskit/osenv_log.c:156
#2 0x1
Laudney Ren <[EMAIL PROTECTED]> writes:
> You are absolutely right!! I ejected my Xircom Cardbus card
> and inserted a 3com non-cardbus card, patched Mach recognized
> it!
Good to hear that the card is recognized, but what would be far
interesting is does it work under Hurd? Could you ping anothe
Laudney Ren <[EMAIL PROTECTED]> writes:
> Then, I applied Wagi's pcmcia patch and when boot up, the same error
> message showedup (refer to my previous email about kernel trap
> error).
Did you boot with a card inserted in a socket? Is this card by chance
a cardbus card? Moritz has send me a bt o
> When calling configure on the OSKit you must ensure that
>
> --enable-indirect-osenv
That's not needed anymore. I spend 3 days wondering why my gnumach
version didn't work.
___
Bug-hurd mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman
Bin Ren <[EMAIL PROTECTED]> writes:
> module /hurd/ext2fs.static
> --multiboot-command-line=${kernel-command-line}
> --host-priv-port=${host-port} --device-master-port=${device-port}
> --exec-server-task=${exec-task} -T typed ${root} $(task-create)
> $(task-resume)
make sure, that the co
> (device driver) EMERGENCY: assertion `FDEV_LINUX_local_irq_count[0] ==0' failed in
>file ../../../../oskit-20020317+pcmcia/linux/dev/softintr.c, line 121 assertion
>`FDEV_LINUX_local_irq_count[0] == 0' failed in file ../../../../oskit-20020317
> +pcmcia/linux/dev/softintr.c, line 121
>
Here some steps how you get pcmcia working. All steps can be done on a
normal GNU/Linux system (no cross compiler needed).
1. Download oskit-20020317 from directly from [1] or better from
Debian [2].
If you downloaded directly from [1], don't forget to apply the Debian
patches (can be do
gnumach and pcmcia status
First, this pcmcia support for gnumach will be only very limited. That
means only for simple ethernet network cards (no multifunctional cards, no
memory cards, etc.)!
Another good news is, that it's seems even possible to insert cards after
booting. Of course, you shoul
> I'm trying to install oskit based Mach with pcmcia patch from wagi.
> Here are my steps and several problems encountered, some of which are
> solved.
oh, a brave soul! :)
> 1. download oskit-20020317.tar.gz from http://www.cs.utah.edu/flux/oskit/
The debian version has some small but important
> I am currently still busy with the console, but this work is nearing to a
> first end. So when I finished that, I can try out my pcmcia card.
Oh, I forgot to mention that up to know I only added my pcmcia network
card. I will add your card today and upload the new patch. So stay
tuned :)
wagi
Good news. The pcmcia patch I made for OSKit seems to work, of course
including some bugs. The patches for OSKit and gnumach are here [1].
I have two major problems found so far. First the hda geometry scan
will _only_ work if the console output goes over the serial line.
Booting without any atta
> What ethernet driver are you using?
The card is ne2000 compatible. The old 8390 driver is used.
During booting the system I see this:
"Use of the PCI-NE2000 driver with this card is recommended!"
Using a newer driver could solve my problem.
wagi
--
eived, 0% packet loss
round-trip min/avg/max = 0.0/2.0/20.0 ms
And also here, the hurd box keeps running.
To conclude (and going to bed) we have certain problems
with oskit-20020317 :(
wagi
--
D
will report these things again on the oskit mailling list. Hopefully
they will vanish by the next release.
wagi
--
Daniel Wagner "Don't drink and derive."
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available o
ecked the return value, instead
I tested for the bytes_written and of course there's only a bogus
value.
wagi
--
Daniel Wagner "Don't drink and derive."
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (p
evice_deallocate(device);
(gdb)
252 if (OutP->RetCode != KERN_SUCCESS)
(gdb)
253 return;
--
Daniel Wagner "Don't drink and derive."
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserv
bytes_written
$1 = (int *) 0xc2bf034
(gdb) p *bytes_written
$2 = 61
wagi
--
Daniel Wagner "Don't drink and derive."
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
Bug-hurd
_write.
I've called device_write with packet size 4, bytes_written was 61 and
the return value was 18677512. Obvious, the semantic is right (packet size
!= written bytes), but I'm not sure if this is ok.
wagi
--
Daniel Wagner "Don't drink and deri
question now is, why
an error code is not a good thing at this location?
wagi
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
Bug
. And therefore the wrong argument does
no harm.
The next step for me is to understand what exactly happends in
the oskit-mach code.
wagi
[1] http://mail.gnu.org/pipermail/bug-hurd/2001-October/005627.html
--
Daniel Wagner "use quit to exit"
I guess that I see something that is 'normal', because I observerd the same
thing without my program running. As you can see, I setted a breackpoint on
ds_device_write and letted the kernel continue. After some time the
breakpoint was hit with the result: kmsg != kmsg->ikm_next->ikm_prev.
wagi
km_next and then comparing with
the value of ikm_prev seems never to match.
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
Bu
any other bugs to.
> I suppose that could somehow be wrong and affect it. Using only
> device_write_inband avoids that code, and so if your problem disappears
> we'll know something.
... so device_write doesn't check correctly all passed parameters?
--
Daniel Wagner
= 0x0, protection = 0x3,
max_protection = 0x7, inheritance = 0x1, wired_count = 0x33c,
user_wired_count = 0x409, projected_on = 0x0}
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1
at ../../oskit-mach/ipc/mach_msg.c:1244
#4 0x0013e253 in mach_call_call ()
#5 0x04113d80 in ?? ()
I know this kind of thing doing to the kernel is nasty, but
still the kernel shouldn't not crash, I think.
Now I'm trying to find out more about this problem and of course how
this can b
The patch removes the not needed assertion.
wagi
2001-11-15 Daniel Wagner <[EMAIL PROTECTED]>
* oskit/osenv_mem.c (free_for_oskit): Don't test for the
OSENV_NONBLOCKING flag. Instead decide on the address which flavor
of memory is to be free
ome memory, so kalloc would fail also, right?
As long I'm not hitting this panic, I'm happy :)
wagi
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
t interrupt level and process it later and it would be hair.
How do I have to alloc memory at this point. I'm not sure what kind
of memory is needed. Is it possible to use zalloc or do have ask lmm for
memory?
wagi
--
Daniel Wagner "use quit to exit"
helped. So it seems definilety to be bug.
> You should definitely post to oskit-users about what that #define is doing
> there. (But note that one of the key guys is out of town right now.)
Ok, I'll post a message about this.
wagi
--
Daniel Wagner &q
IC);
+ skb = kmalloc(SKB_HDRSIZE + size, GFP_ATOMIC | mflags);
gotone:
linux_restore_flags(flags);
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (p
On Mon, 05 Nov 2001, Daniel Wagner wrote:
> On Mon, 29 Oct 2001, Daniel Wagner wrote:
>
>
> Fixing kmalloc was easy, but I've some problems with kfree. kfree takes
Oops, it should be oskit_skbufio_mem_alloc and not kmalloc. Here is the
bt for freeing memory:
#0 panic (
On Mon, 29 Oct 2001, Daniel Wagner wrote:
> oskit and not for oskit-mach. So I will write a patch for oskit and then
> I'll see what happends.
It seems that the oskit wrappers for kmalloc and kfree don't pass the
right flags to oskit_linux_mem_alloc and oskit_linux_mem_free.
, but the kernel would still crash on this test, because the
assertion is not true. Of course a prober patch would be for
oskit and not for oskit-mach. So I will write a patch for oskit and then
I'll see what happends.
wagi
--
Daniel Wagner "use quit to
if (err)
error (2, err "%s", interface);
mach_port_deallocate (mach_task_self (), ether_port);
ports_destroy_right (readptname);
return 0;
}
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (publ
Hi
The following patch dissables a test on the OSENV_NONBLOCKING flag.
This is needed because there is a discrepancy between the oskit
docs and the code. Though the docs says it is set, it is not. This
prevents oskit-mach from correct working.
wagi
2001-10-28 Daniel Wagner <[EM
what?
wagi
--
Daniel Wagner "use quit to exit"
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
Bug-hurd mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-hurd
= config.h cpus.h
headerfiles = $(filter-out $(nr_headerfiles),$(wildcard *.h))
clean:
rm -f *.[oda] kernel
rm -f *.symc *.symc.o *_user.c *_server.c $(headerfiles)
rm -f *.migs_d *.migsh_d *.migu_d *.miguh_d
wagi
--
Daniel Wagner
It's still working! And it's much more faster than before, since there are no
blockings in the traphandler anymore. I hope that's still correct. I also changed
all enable/disable calls from intr to softintr.
> I'm glad to hear it's working for you, but that's still not quite the way
> I'd like t
Good news. The applied patch seems to work correctly. I tested oskit-mach
for about an hour with a stress test. No panics:)
Please look through the patch, and tell me how far away from a good patch
I am.
wagi
Index: osenv_timer.c
==
rrectly, because they did also counting down a counter. The result was
used as a dividend which caused and division by zero. AMD removed this
optimation later, since most people thought it was AMD fault and not
MS... It might be the same here.
wagi
--
Daniel Wagner
just call setsoftclock after setting your mask (at
> splsoftclock).
Ok, I'll change it according your tips.
wagi
--
Daniel Wagner
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
Bug-hurd
Ok, here is my first attempt. White the patch below oskit-mach should
work. Unfortunately there is a (some?) bug(s). Under heavy net load
oskit-mach traps after some time. Maybe someone know what I have done wrong.
wagi
Index: pc/osenv_timer.c
==
Finaly, I managed to it get working! Roland, your tips were excactly
what I needed! Many thanks. I will send a patch after I have clean up
the new code. Right now it's just a big hack, but still it works :)
cheers,
wagi
--
Daniel Wagner
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (p
frontend :)
> Also, pardon my ignorance :-) but what's the command to log a GDB
> session to a file?
script gdb ...
wagi
--
Daniel Wagner
email: [EMAIL PROTECTED]
GnuPG: 1024D/DCDE890A (public key available on any keyserver)
___
Bug-hur
an take a look at how they use the PGE feature and see if
> it differs from what I did in oskit-mach.
There's not much difference, as I can see. Differnent names but it is used in
the same way. Maybe I am wrong here. Someone with more expirence might have a
look at it.
wagi
--
Daniel Wag
the
> PGE extension (judging from the cpuid output), the code was trying to enable
> the flag for this feature in the %cr4 register and that generated the GPF.
It seems that I have the same problem, but I've got an K6-2. So that might be K6
family error.
wagi
--
Daniel Wagner
email: [EMA
63 matches
Mail list logo
