Hi list,
The current 8139too.c linux kernel driver dosn't work.
There is the patch for the driver 8139too.c at :
http://descript.sysdoor.net/patch/kernel/2.4.19/8139too.c.diff
It fix some problems with card mode, new hard detection and new card
added.
Please read the diff.
PS : There is something strange with the list... or perhaps me ...
Best regards,
Vergoz Michael
Vergoz Michael wrote:
> Hi list,
>
> The current 8139too.c linux kernel driver dosn't work.
Please be more specific.
> There is the patch for the driver 8139too.c at :
>
> http://descript.sysdoor.net/patch/kernel/2.4.19/8139too.c.diff
>
> It fix some problems with card mode, new hard detection and new card
> added.
>
> Please read the diff.
The diff is huge, mostly unnecessary, and backs out obvious bug fixes
(i.e. it _adds_ bugs). It removes several PCI IDs, eliminating 8139
support for some cards. The "new card" is supported by another driver,
8139cp.c.
Please send _specific_ changes and bug fixes.
Jeff
Hi Jeff,
What i see is the current driver _doesn't_ work on my realtek 8139C.
With this one it work fine.
Regards,
Michael
Sent: Monday, November 18, 2002 8:34 AM
Subject: Re: 8139too.c patch for kernel 2.4.19
> Vergoz Michael wrote:
>
> > Hi list,
> >
> > The current 8139too.c linux kernel driver dosn't work.
>
>
> Please be more specific.
>
> > There is the patch for the driver 8139too.c at :
> >
> > http://descript.sysdoor.net/patch/kernel/2.4.19/8139too.c.diff
> >
> > It fix some problems with card mode, new hard detection and new card
> > added.
> >
> > Please read the diff.
>
>
>
> The diff is huge, mostly unnecessary, and backs out obvious bug fixes
> (i.e. it _adds_ bugs). It removes several PCI IDs, eliminating 8139
> support for some cards. The "new card" is supported by another driver,
> 8139cp.c.
>
> Please send _specific_ changes and bug fixes.
>
> Jeff
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
Vergoz Michael wrote:
> Hi Jeff,
>
> What i see is the current driver _doesn't_ work on my realtek 8139C.
> With this one it work fine.
What does not work? Does the driver load? Can you ping? Does the
driver work for a little while then stop? Does it work perfectly except
under heavy load?
Also, please read "REPORTING-BUGS" file in a Linux kernel source tree
for additional debugging information to provide (lspci, dmesg, etc.)
Your patch includes so much junk it is obviously unacceptable.
Therefore, I request that you work with me (and other kernel developers)
to narrow down the problem, so that it may be fixed.
Regards,
Jeff
Jeff,
just as a hint: I do have a SiS based UP test system around that
misbehaves with IO-APIC enabled.
See
http://www.msi.com.tw/program/products/slim_pc/slm/pro_slm_detail.php?UID=134&MODEL=MS-6232
for more system details.
From the back of my head: with IO-APIC enabled and 3 RTL8139 (one on
board, the other two PCI) the NICs get assigned IRQs 17 and 18 but don't
seem to receive any interrupt at all. Without IO-APIC all three NICs
share IRQ 11 and do work. Further details on request (very low priority
problem for me). Oh, nearly forgot: kernel is 2.4.20rc1.
--
Andreas Steinmetz
D.O.M. Datenverarbeitung GmbH
Andreas Steinmetz wrote:
> Jeff,
> just as a hint: I do have a SiS based UP test system around that
> misbehaves with IO-APIC enabled.
> See
> http://www.msi.com.tw/program/products/slim_pc/slm/pro_slm_detail.php?UID=134&MODEL=MS-6232
>
> for more system details.
> From the back of my head: with IO-APIC enabled and 3 RTL8139 (one on
> board, the other two PCI) the NICs get assigned IRQs 17 and 18 but don't
> seem to receive any interrupt at all. Without IO-APIC all three NICs
> share IRQ 11 and do work. Further details on request (very low priority
> problem for me). Oh, nearly forgot: kernel is 2.4.20rc1.
That's not going to be fixed by Michael's patch... Any IOAPIC-related
problems cannot be fixed at the driver level, but must be fixed by a
BIOS update (or possibly an IOAPIC code fix). Sometimes vendors do not
bother do even wire the IOAPIC when it is a uniprocessor board :(
Jeff
Jeff Garzik wrote:
> That's not going to be fixed by Michael's patch... Any IOAPIC-related
> problems cannot be fixed at the driver level, but must be fixed by a
> BIOS update (or possibly an IOAPIC code fix). Sometimes vendors do not
> bother do even wire the IOAPIC when it is a uniprocessor board :(
I just wanted to point exactly in that (IO-APIC) direction as a common
source for trouble.
--
Andreas Steinmetz
D.O.M. Datenverarbeitung GmbH
Andreas Steinmetz wrote:
> Jeff Garzik wrote:
>
> > That's not going to be fixed by Michael's patch... Any IOAPIC-related
> > problems cannot be fixed at the driver level, but must be fixed by a
> > BIOS update (or possibly an IOAPIC code fix). Sometimes vendors do
> > not bother do even wire the IOAPIC when it is a uniprocessor board :(
>
>
> I just wanted to point exactly in that (IO-APIC) direction as a common
> source for trouble.
Oh agreed. A very annoying, common source of trouble :)
On Mon, Nov 18, 2002 at 09:50:50AM +0100, Vergoz Michael wrote:
> Hi Jeff,
>
> What i see is the current driver _doesn't_ work on my realtek 8139C.
> With this one it work fine.
The question was (as I understand the changes you made and the
answer from Jeff) : did your card work with 8139cp or not?
If not, you have to modify 8139cp, which is the right place for C+ support.
--
Ducrot Bruno
http://www.poupinou.org Page profaissionelle
http://toto.tu-me-saoules.com Haume page
Vergoz Michael wrote:
> Hi Jeff,
>
> What i see is the current driver _doesn't_ work on my realtek 8139C.
> With this one it work fine.
What you have send to the list is a huge change:
ShuChen Shao took the 0.9.15 driver and added 8139 C+ support to it. This will not go into the kernel, instead the kernel contains a new driver for the 8139C+ nic.
You took his driver and added additional changes to it.
Meanwile development continued in the kernel driver [version 0.9.25].
Now we must try to merge everything again. Unfortunately this means manually figuring out which of your changes are required and which back out bugfixes.
Could you try to shrink your patch down to the necessary changes?
- did the 8139c+ driver from ShuChen work?
- create a diff with just your changes.
And add some description about your problem.
--
Manfred
Ducrot Bruno wrote:
> On Mon, Nov 18, 2002 at 09:50:50AM +0100, Vergoz Michael wrote:
>
> >Hi Jeff,
> >
> >What i see is the current driver _doesn't_ work on my realtek 8139C.
> >With this one it work fine.
>
>
> The question was (as I understand the changes you made and the
> answer from Jeff) : did your card work with 8139cp or not?
>
> If not, you have to modify 8139cp, which is the right place for C+
> support.
Agreed. However from the above quoted, "8139C" chip would indicate that
he needs to use 8139too not 8139cp.
I am hoping (please!) that Michael will post some info that will help us
debug his problem.
Regards,
Jeff
Hi list,
The driver that i have post support realtek 8139A/B/C/D/C+ (already tested)
and another dlink card (realtek based).
On certain motherboard where the chip is integred you will perhaps have some
troubles.
The driver 8139cp.c can be removed to the source entry.
I think it's the better way to removed it, but i'm not a linux kernel
developer to take any type of decisions :P
PS : I'v don't really understand what is the problem with the APIC support
?!
Best regards,
Vergoz Michael
From: "Jeff Garzik" <[email protected]>
To: "Ducrot Bruno" <[email protected]>
Cc: "Vergoz Michael" <[email protected]>; "Linux Kernel Mailing List"
<[email protected]>
Sent: Monday, November 18, 2002 6:25 PM
Subject: Re: 8139too.c patch for kernel 2.4.19
> Ducrot Bruno wrote:
>
> > On Mon, Nov 18, 2002 at 09:50:50AM +0100, Vergoz Michael wrote:
> >
> > >Hi Jeff,
> > >
> > >What i see is the current driver _doesn't_ work on my realtek 8139C.
> > >With this one it work fine.
> >
> >
> > The question was (as I understand the changes you made and the
> > answer from Jeff) : did your card work with 8139cp or not?
> >
> > If not, you have to modify 8139cp, which is the right place for C+
> > support.
>
>
>
> Agreed. However from the above quoted, "8139C" chip would indicate that
> he needs to use 8139too not 8139cp.
>
> I am hoping (please!) that Michael will post some info that will help us
> debug his problem.
>
> Regards,
>
> Jeff
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
Vergoz Michael wrote:
> Hi list,
>
> The driver that i have post support realtek 8139A/B/C/D/C+ (already
> tested)
> and another dlink card (realtek based).
> On certain motherboard where the chip is integred you will perhaps
> have some
> troubles.
Would you be willing to work with Linux kernel developers to integrate
the useful changes in this driver into the kernel?
The driver you posted is my driver, version 0.9.15, which is an older
version of the kernel's driver and doesn't include many changes made
since then.
If you are seeing a problem, please describe in detail the problem so
that kernel developers may fix it in the current kernel's 8139too.c.
> The driver 8139cp.c can be removed to the source entry.
This is the preferred driver for the 8139C+ chip, as it included many
bug fixes and is much faster than the version from RealTek.
Jeff
> Vergoz Michael wrote:
>
> > Hi list,
> >
> > The driver that i have post support realtek 8139A/B/C/D/C+ (already
> > tested)
> > and another dlink card (realtek based).
> > On certain motherboard where the chip is integred you will perhaps
> > have some
> > troubles.
>
>
> Would you be willing to work with Linux kernel developers to integrate
> the useful changes in this driver into the kernel?
* of course
>
> The driver you posted is my driver, version 0.9.15, which is an older
> version of the kernel's driver and doesn't include many changes made
> since then.
* Yes, it's yours, but it work :P
>
> If you are seeing a problem, please describe in detail the problem so
> that kernel developers may fix it in the current kernel's 8139too.c.
* The problem is in packet recv, i'v to work this night to find what exactly
is the problem
>
> > The driver 8139cp.c can be removed to the source entry.
>
>
> This is the preferred driver for the 8139C+ chip, as it included many
> bug fixes and is much faster than the version from RealTek.
* I will make a prupose to all linux kernel developers about that. I prefere
to fix that problem before to prupose anything else :P
>
> Jeff
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
Best regards,
Vergoz Michael
Vergoz Michael wrote:
> >Vergoz Michael wrote:
> >If you are seeing a problem, please describe in detail the problem so
> >that kernel developers may fix it in the current kernel's 8139too.c.
>
>
> * The problem is in packet recv, i'v to work this night to find what
> exactly
> is the problem
thanks!
>
>
> >>The driver 8139cp.c can be removed to the source entry.
> >
> >
> >This is the preferred driver for the 8139C+ chip, as it included many
> >bug fixes and is much faster than the version from RealTek.
>
>
> * I will make a prupose to all linux kernel developers about that. I
> prefere
> to fix that problem before to prupose anything else :P
Certainly. Comments from all are welcome. Here is the general argument:
8139/A/B/C/D, 8100, 8139 rev K. are all supported by 8139too.
8139C+ is a totally different chip, and thus gets a totally different
driver. The realtek 8139C+ changes (which you posted) basically add a
totally different driver into the same source code. The only reason for
this is that PCI ID for 8139C+ is similar [differs on PCI Revision Id].
finally, I emailed my RealTek contact directly, and he said RealTek has
found no problems with 8139cp.c driver, and they support 8139cp's use.
Regards,
Jeff
Jeff Garzik wrote:
> 8139C+ is a totally different chip, and thus gets a totally different
> driver.
Well, to correct myself before anyone calls me on it, 8139too and 8139cp
do share some phy code.
However 8139cp phy code is obviously simplified from 8139too since it
does not support external phys in the driver like 8139too does with the
"8129" chip.
Jeff