etherboot-developers — Discussion of developer issues

[Etherboot-developers] Re: [ANNOUNCE] mkelfImage 2.0

[ollie l. <ol...@si...>]

On Tue, 2003-01-14 at 08:19, Eric W. Biederman wrote:
> After a long and tedious grind I am pleased to announce version 2.0 of mkelfImage.
> New features.
> - One single binary
> - Itanium support
> - All of the code is in C.
> - New and exciting easter eggs (err bugs can you find them before I do?...)
> 
> 
> Available at: ftp://ftp.lnxi.com/pub/src/mkelfImage/mkelfImage-2.0.tar.gz
> 
> rpms can be built with rpm -ta mkelfImage-2.0.tar.gz
> alien will make perfectly serviceable debs from those...
> 
> The decision to go with pure C is that it is both relatively portable,
> and it reduces maintenance when I add a feature.  I tried perl and it
> handles the interesting code only with the utmost protest.  Adding
> checksums and other interesting pieces is much easier if it only needs
> to be both coded in one language, and usually the C implementation is
> much easier than the perl.  Plus C is fast :)

Eric,
	Something off topic, the elfImage made by mkelfImage 1.1x can 
not be recognized by 'objdump' but can be dumped by 'readelf'. Is there
any reason for this ? Is this fixed by ver. 2.0 ?
 
-- 
ollie lho <ol...@si...>

[Etherboot-developers] [ANNOUNCE] mkelfImage 2.0

[<ebi...@ln...>]

After a long and tedious grind I am pleased to announce version 2.0 of mkelfImage.
New features.
- One single binary
- Itanium support
- All of the code is in C.
- New and exciting easter eggs (err bugs can you find them before I do?...)


Available at: ftp://ftp.lnxi.com/pub/src/mkelfImage/mkelfImage-2.0.tar.gz

rpms can be built with rpm -ta mkelfImage-2.0.tar.gz
alien will make perfectly serviceable debs from those...

The decision to go with pure C is that it is both relatively portable,
and it reduces maintenance when I add a feature.  I tried perl and it
handles the interesting code only with the utmost protest.  Adding
checksums and other interesting pieces is much easier if it only needs
to be both coded in one language, and usually the C implementation is
much easier than the perl.  Plus C is fast :)

Have fun,

Eric

Re: [Etherboot-developers] Checksum Failed

[<ebi...@ln...>]

ke...@us... (Ken Yap) writes:

> >Additionally there has been some research in how to implement trusted
> >computing.  Which I believe means verifying an image signature when
> >the image is downloaded.  And with an infrastructure to check an image
> >checksum it is clear where to put the hooks to verify an image
> >signature.
> 
> Yes, speculating if Etherboot could support USB, one thing you could do
> is use a USB dongle to obtain a key with which one could verify a signed
> image. Or maybe the key could be held in the ROM.

And the relevant link to that work is:
http://www.missl.cs.umd.edu/Projects/sebos/main.shtml

There has been activity but I have not seen the contributing
back patches yet..  Though I suspect this would be very useful
on a wireless network or a wide area network.

Eric

Re: Re: [Etherboot-developers] isapnp

[Timothy L. <tl...@ro...>]

> Now USB and PCMCIA, those are more interesting targets than the 3c509,
> nudge nudge.

You are correct of course.  I have purchased a USB controller for my development machine.  I don't mind looking at USB, but what ever I come up with will probably be a somewhat nasty looking hack of the Linux code. 

I am about to admit defeat on the Sundance card (I cannot find the problem that makes it so slow) so I have time to either rewrite isapnp or look at USB.  Of course, I could probably rewrite isapnp in a few days, USB is an entirely different story.

Tim 

Re: [Etherboot-developers] isapnp

[<ke...@us...>]

>If I remember correctly, there is a section in the code that basically
>gives up after finding the 3c515.  I assume that you changed that for
>your specific nic.  I have been going to try the code with the 3c509.
>Most Linux documentation says to disable PNP, but that may have been the
>recommendation before ISAPNP was well implemented.

I noted that the CS8920 PNP IDs were detected and left it at that. I
won't be doing any more work on the CS8920, it was just a lark. These
machines already have an EEPRO100 NIC in them in addition to the onboard
10 Mb NIC, probably put in when the network was upgraded to 100 Mb.
There's no incentive to avoid using the 100 Mb NICs. Especially when one
can get new 100 Mb NICs so cheap now.

Now USB and PCMCIA, those are more interesting targets than the 3c509,
nudge nudge.

RE: [Etherboot-developers] isapnp

[Timothy L. <tl...@ro...>]

> Unfortunately it won't be any use for now, because the Linux driver
> doesn't support PnP anyway, there a comment to the effect "please turn
> off PnP".
> 
> It would be interesting if experimenters could try isapnp.c with other
> ISA PnP cards. Then again, ISA cards are disappearing and most of them
> can have PnP disabled anyway.

If I remember correctly, there is a section in the code that basically
gives up after finding the 3c515.  I assume that you changed that for
your specific nic.  I have been going to try the code with the 3c509.
Most Linux documentation says to disable PNP, but that may have been the
recommendation before ISAPNP was well implemented.

Tim

RE: [Etherboot-developers] isapnp

[Timothy L. <tim...@us...>]

> kusn> Unfortunately it won't be any use for now, because the Linux
driver
> kusn> doesn't support PnP anyway, there a comment to the effect
"please
> turn
> kusn> off PnP".
> 
> I'm not sure about it, but could etherboot assign the card an io/irq
> combination so that it is initialized /so far/ when entering Linux
> boot?

This is one of the areas that troubled me.  I did not know how to get a
free irq or dma to assign to the card.  In the case of the 3c515, it
does not seem to care so I did not assign it an irq or dma (dma might be
incorrect; my memory is several months old).  

Tim

RE: [Etherboot-developers] 5.1 will not boot on 486

[Timothy L. <tim...@us...>]

> I still think it's the printf itself that's blocking after printing
one
> character. I still think the serial port is being written to somehow.
> Even though the serial port may be present, if the handshake lines are
> not enabled, the write to the port may block.
> 
> What happens if you put another printf ahead the current one, this
time
> with a different first character?

I thought that as well.  I wanted to see what was going on, so I put a
single printf before the current one.  That printf also failed after
outputting the first character.  I followed up with an infinite while
loop to printf a word.  That also failed after printing the first
character of the first iteration.

Tim

Re: [Etherboot-developers] isapnp

[Anselm M. H. <an...@ho...>]

Hello ken,

kusn> Been playing a bit with the standalone isapnp module Timothy Legge just
kusn> hacked. There's this old IBM machine with an onboard CS8920 here. After
kusn> a couple of hacks to start the scan at 0x203 and to slow down the
kusn> checksum bit acquisition algorithm, I can get it to detect the CS8920
kusn> entry.

kusn> Unfortunately it won't be any use for now, because the Linux driver
kusn> doesn't support PnP anyway, there a comment to the effect "please turn
kusn> off PnP".

I'm not sure about it, but could etherboot assign the card an io/irq
combination so that it is initialized /so far/ when entering Linux
boot?

Best regards,
 Anselm                            mailto:an...@ho...

[Etherboot-developers] 5.1.5+: WD80x3 driver working and CVSed

[<ke...@us...>]

Finally got the bus_to_virt thing sorted out. Finishes loading, says
done, but jumps into hyperspace. May not be a driver problem now.

Re: [Etherboot-developers] 5.1 will not boot on 486

[<ke...@us...>]

>I verified that the 386 option is being used, or at least I think I did.
>I assume it just needs to be added in the CFLAGS section of the Config
>files.  
>
>I am assuming that the issue occurs somewhere between the printing of
>done in the boot1a.S file and the main function of main.c.
>
>What if anything happens between the two events?  

I don't think it's a code generation issue anyway, those are usually
fatal quite soon.

I still think it's the printf itself that's blocking after printing one
character. I still think the serial port is being written to somehow.
Even though the serial port may be present, if the handshake lines are
not enabled, the write to the port may block.

What happens if you put another printf ahead the current one, this time
with a different first character?

[Etherboot-developers] isapnp

[<ke...@us...>]

Been playing a bit with the standalone isapnp module Timothy Legge just
hacked. There's this old IBM machine with an onboard CS8920 here. After
a couple of hacks to start the scan at 0x203 and to slow down the
checksum bit acquisition algorithm, I can get it to detect the CS8920
entry.

Unfortunately it won't be any use for now, because the Linux driver
doesn't support PnP anyway, there a comment to the effect "please turn
off PnP".

It would be interesting if experimenters could try isapnp.c with other
ISA PnP cards. Then again, ISA cards are disappearing and most of them
can have PnP disabled anyway.

RE: [Etherboot-developers] 5.1 will not boot on 486

[Timothy L. <tim...@us...>]

> >> Just a thought, check that gcc is set to generate 386 code in all
> >> Makefiles, except the Itanium directories of course
> >
> >Thanks for the information.  What exactly am I looking for?
> 
> That gcc is being called with the -mcpu=i386 option.

I verified that the 386 option is being used, or at least I think I did.
I assume it just needs to be added in the CFLAGS section of the Config
files.  

I am assuming that the issue occurs somewhere between the printing of
done in the boot1a.S file and the main function of main.c.

What if anything happens between the two events?  

Tim

Re: [Etherboot-developers] Geode + Etherboot + Syslinux + DiskOnChip + RTL8139C help

[<ke...@us...>]

>ROM segment 0x1000 length 0x4000 reloc 0x8400
>A20 enabled via BIOS
>Etherboot 5.0.8 (GPL) Tagged ELF for [RTL8139]
>Boot from (N)etwork or from (L)ocal? N
>Probing...[RTL8139]pci_init: no BIOS32 detected
>No adapter found
><sleep>
><abort>
>Boot from (N)etwork or from (L)ocal?

No BIOS32 means that your BIOS isn't a standard PCI PnP BIOS which
Etherboot can ask for the details of the devices on the bus. This is
often the case with embedded BIOSes which don't try to emulate desktop
BIOSes. You may have to figure out how to get that PCI information.

[Etherboot-developers] Geode + Etherboot + Syslinux + DiskOnChip + RTL8139C help

[Ralph B. <cy...@hq...>]

Hello,

   Im having a hard time with Etherboot im hoping someone on this list can help
me with :)  I have a VisionBank ST320 (aka Jammin-125 compatible machine) im
trying to setup as a LTSP terminal, but I cannot get it to see the RTL8139C
network chip.  I am using a 2 meg DiskOnChip 2000 with Syslinux 2.00 installed
on it.  (DOS FAT filesystem)  Bios is an XpressROM V3.2.2 Rev BNI01.05

After doing extensive google searches, i have have read about the
Etherboot/syslinux/diskonchip memory overlap issues.  I have set RELOCADDR to
0x84000 (along with others during troubleshooting) but i continually get the
following error message:

----------------------------------------------------
SYSLINUX 2.00 .............
boot:
Loading linux.....Ready.
ROM segment 0x1000 length 0x4000 reloc 0x8400
A20 enabled via BIOS
Etherboot 5.0.8 (GPL) Tagged ELF for [RTL8139]
Boot from (N)etwork or from (L)ocal? N
Probing...[RTL8139]pci_init: no BIOS32 detected
No adapter found
<sleep>
<abort>
Boot from (N)etwork or from (L)ocal?
----------------------------------------------------

Any ideas?  I tried to figure this one out on my own, learning a lot about
etherboot in the process, but I just cant get it to see the card.  (linuxbios
rocks - i cant wait to get it on this machine someday)

I know the interface works.  If i take out the DiskOnChip and insert a WinCE
32meg IDE Flash drive, I can access the internet with the built in, but quite
useless, IE browser :)

-- 
--^-^--^-^-^---^--^-^--^-^-^--^---^-^--^-^-^--^-^--^-^-^--
 Ralph Bonnell @ Linux Friendly, Inc.
 - CISSP, LPIC-2, CCSI, CCSE+, CCNA, CSFE, MCSE 2000
 +1-954-931-7388   http://ralph.cx   ra...@ra...
--^-^--^-^-^---^--^-^--^-^-^--^---^-^--^-^-^--^-^--^-^-^--

Re: [Etherboot-developers] 5.1 will not boot on 486

[<ke...@us...>]

>> Just a thought, check that gcc is set to generate 386 code in all
>> Makefiles, except the Itanium directories of course
>
>Thanks for the information.  What exactly am I looking for?

That gcc is being called with the -mcpu=i386 option.

RE: [Etherboot-developers] 5.1 will not boot on 486

[Timothy L. <tl...@ro...>]

> Just a thought, check that gcc is set to generate 386 code in all
> Makefiles, except the Itanium directories of course

Thanks for the information.  What exactly am I looking for?

Re: [Etherboot-developers] 5.1 will not boot on 486

[<ke...@us...>]

>I actually commented out everything in main() before the printf and it
>failed in the same place.

Just a thought, check that gcc is set to generate 386 code in all
Makefiles, except the Itanium directories of course

Re: [Etherboot-developers] Checksum Failed

[<ke...@us...>]

>Additionally there has been some research in how to implement trusted
>computing.  Which I believe means verifying an image signature when
>the image is downloaded.  And with an infrastructure to check an image
>checksum it is clear where to put the hooks to verify an image
>signature.

Yes, speculating if Etherboot could support USB, one thing you could do
is use a USB dongle to obtain a key with which one could verify a signed
image. Or maybe the key could be held in the ROM.

[Etherboot-developers] Etherboot 5.1.5 (development) released

[<ke...@us...>]

I have released Etherboot 5.1.5 at
http://www.etherboot.org/distribution.html.

There are more fixes to Itanium support and the drivers. Please test it
and post feedback to this mailing list. Quick reminder: it's

	make bin/3c509.zrom

not

	make bin32/3c509.lzrom

now, similarly for other .lz targets.

Re: [Etherboot-developers] Checksum Failed

[<ebi...@ln...>]

"Timothy Legge" <tl...@ro...> writes:

> > In this case it sounds like it may have been triggered by a sundance
> > bug.
> > 
> In this particular case, I ifdefed out an unneeded code section and the
> error went away.  Could the image produced have been too big?

Unlikely.   There is nothing in the code base that actually cares
about the image size.   There might be an alignment issue somewhere
though.  If the problem is reproducible it should be possible to track
it down.  Is the problem reproducible?


Eric

[Etherboot-developers] [±¡°í]ÀλýÀÌ Áñ°Å¿ö Áý´Ï´Ù!!!

[mail<lif...@em...>]

<HTML>
<a href=http://www.madpia.com/index.asp?r_id=mail>
<img src=http://www.madpia.com/banner/banner2.gif border=0></a>
<BR><FONT size=2>안녕하세요? mail입니다.
<BR>좋은 정보로 부자되세요!!!
<BR>매드피아는 영화을 상영하는 인터넷 종합 영화사이트 입니다.
<BR>업데이트되는 각종 영화를 마음껏 보시면서 회원가입만으로 7단계 완성시
<BR>매월 229만원 까지의 금액을 받을 수 있고, 추가적으로 3명만 추천가입시키면
<BR>매월 6천2백만원까지 받을 수 있도록 설계되어 있습니다.
<BR>인생이 즐거워 집니다!!!</FONT>
<BR><a href=http://www.madpia.com/index.asp?r_id=mail>
<img src=http://www.madpia.com/banner/banner1.gif border=0></a>
<BR><FONT color=blue size=2>추천인ID: mail </FONT>
<BR><FONT size=2>-3*10 매트릭스 구성, 회원유치시 7,000원, 매월 &nbsp;700원씩&nbsp;</FONT>
<FONT size=2><BR>-자신의 추천인이 3명 이상일 경우, 4번째 추천인은 3명 아래로
<BR>-물론 추천금 7,000원 본인에게
<BR>-</FONT><FONT color=blue size=2>유료인거만큼 수익성이 아주 확실합니다.
<BR>-하부 회원은 자신이 하지 않아도 자동으로 계속 생성됩니다.
<BR>-상위회원의 추천인이 본인의 라인으로 되는 것입니다.
<BR>-업데이트되는 최신 영화를 마음껏 보기 때문에 더욱 확실합니다.
<BR>-무료로 제공되는 대용량 데이터 1GB의 웹 저장공간을 통해
<BR>각종 자료를 접속만 하면 회원 누구나 저장/이동/공유가 가능합니다.
<BR>-매드배팅은 매달 지급해 드리는 충전매드로 베팅게임을 즐기실 수 있습니다.
<BR>스포츠 베팅 서비스로 흥미만점!  베팅게임도 즐기고 매드도 벌어 보세요.
<BR>그럼, 좋은 하루 보내세요!   </FONT>
<BR><a href=http://www.madpia.com/index.asp?r_id=mail>
<img src=http://www.madpia.com/banner/banner3.gif border=0></a>
<BR><BR><BR><A href="mailto:jun...@ko...">수신거부</A></HTML>

Re: [Etherboot-developers] Checksum Failed

[<ebi...@ln...>]

"Timothy Legge" <tl...@ro...> writes:

> Hi
> 
> I was cleaning up the sundance driver a little.  When I compiled the
> driver, I received a checksum error and I get a Boot Err when booting
> the client.
> 
> What is the check sum error?

The latest version of mkelf computes a simple ip checksum over
the entire loaded image.  The etherboot elf loader checks that
checksum (only if it is present in the ELF note section) and if it
does not match you see an error loading the image.

In this case it sounds like it may have been triggered by a sundance
bug.

The original motivation was that occasionally with bad hardware I see
a corrupted data transfer.  A checksum on the image allows me to
see if it was loaded across the network correctly.  

Additionally there has been some research in how to implement trusted
computing.  Which I believe means verifying an image signature when
the image is downloaded.  And with an infrastructure to check an image
checksum it is clear where to put the hooks to verify an image
signature.

Eric

[Etherboot-developers] Checksum Failed

[Timothy L. <tl...@ro...>]

Hi

I was cleaning up the sundance driver a little.  When I compiled the
driver, I received a checksum error and I get a Boot Err when booting
the client.

What is the check sum error?

Tim

[Etherboot-developers] Booting a LILO directly from Etherboot

[William H. F. <wil...@sb...>]

I am using Etherboot on a tiny scratch drive (IDE2, master).  Interestingly,
the BIOS standard is not to boot twice from the same class of device
(HD->ROM->removable,etc) is OK, but HD->HD is not.  So in order to do the
local boot, I modified the sources to boot the Linux disk themselves.
However, I get the 'L' followed by about 256 error 99s.  I am investigating
why this happens and thought I would ask the group if anyone had knowledge
of this.  I'm new here and hate to rush in with a posted question.  If I
have committed a foopah, I apologize in advance.

Cheers