[ipxe-devel] ipxe infiniband layer possible bug
Alex Markuze
markuze at mellanox.com
Fri Nov 8 13:39:44 UTC 2013
Hi, Milan.
The ConnectX3 cards have Mellanox modified version of ipxe , FlexBoot(http://www.mellanox.com/page/products_dyn?product_family=34&mtag=flexboot).
The issue you are describing may be connected to autosensing , when the nic doesn't recognize the link layer correctly.
Please try burning the latest version of FlexBoot, see if that helps.
You should contact your vendor and receive support for the issue you are seeing.
-----Original Message-----
From: ipxe-devel-bounces at lists.ipxe.org [mailto:ipxe-devel-bounces at lists.ipxe.org] On Behalf Of ipxe-devel-request at lists.ipxe.org
Sent: Friday, November 08, 2013 2:00 PM
To: ipxe-devel at lists.ipxe.org
Subject: ipxe-devel Digest, Vol 42, Issue 9
Send ipxe-devel mailing list submissions to
ipxe-devel at lists.ipxe.org
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.ipxe.org/mailman/listinfo.cgi/ipxe-devel
or, via email, send a message with subject or body 'help' to
ipxe-devel-request at lists.ipxe.org
You can reach the person managing the list at
ipxe-devel-owner at lists.ipxe.org
When replying, please edit your Subject line so it is more specific than "Re: Contents of ipxe-devel digest..."
Today's Topics:
1. ipxe infiniband layer possible bug (simi77 at email.cz)
2. Re: suggestion for improvement - USB Logging/Debug
(Petry Roman, SHS-S IT-TS)
3. Re: undi interface info (Larry Brigman)
4. Re: suggestion for improvement - USB Logging/Debug (Michael Brown)
5. Re: suggestion for improvement - USB Logging/Debug
(Petry Roman, SHS-S IT-TS)
----------------------------------------------------------------------
Message: 1
Date: Wed, 06 Nov 2013 12:41:46 +0100 (CET)
From: <simi77 at email.cz>
To: <ipxe-devel at lists.ipxe.org>
Subject: [ipxe-devel] ipxe infiniband layer possible bug
Message-ID: <2L{.{OnR.280ELEuZj6.1IUYhw at seznam.cz>
Content-Type: text/plain; charset="us-ascii"
Hello dears,
I'm testing latest (few days ago) git version iPXE trying to boot over IB.
MT26428 Hermon in client, QL/ST12800 fabrick, OpenSM or Topspin TS270 SM . Link starts in DOWN state, driver activates INIT state , then MI debug (ib_mi) is printed on screen - TX, RX status 0000.
After number of tryes iPXE gives up. Link never comes into ARMED or ACTIVE state. When booting Linux over Eth. and mlx4 driver activates link, IB if fully functional. Does somebody have any idea ?
THX.
Sincerely
Milan Simanek
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ipxe.org/pipermail/ipxe-devel/attachments/20131106/87ac5711/attachment-0001.html>
------------------------------
Message: 2
Date: Thu, 7 Nov 2013 18:21:33 +0100
From: "Petry Roman, SHS-S IT-TS" <Roman.Petry at shsservices.org>
To: "Michael Brown" <mbrown at fensystems.co.uk>
Cc: ipxe-devel at lists.ipxe.org
Subject: Re: [ipxe-devel] suggestion for improvement - USB
Logging/Debug
Message-ID:
<116C0EBA85E1C64DA9891D284A992E8B854539 at win84.dh-com.continental.steel>
Content-Type: text/plain; charset="iso-8859-1"
8-)...
Nice try to get such a device 8-)..
if i understand the linux ehci gadget debug module rught, itwill make your normal linux pc to a compatible debug device..
so just take your pc or raspberry pi and connect it via usb to the normal legacy pc you want to debug, and it should work..
Bye
roman
-----Urspr?ngliche Nachricht-----
Von: Michael Brown [mailto:mbrown at fensystems.co.uk]
Gesendet: Donnerstag, 7. November 2013 17:12
An: Petry Roman, SHS-S IT-TS
Cc: ipxe-devel at lists.ipxe.org
Betreff: Re: [ipxe-devel] suggestion for improvement - USB Logging/Debug
On 06/11/13 15:47, Petry Roman, SHS-S IT-TS wrote:
> Hello, just a suggestion for a feature to implement: EHCI Debug Port
>
> As more and more pc?s are legacy free.. so no serial port for debuging,
> this could be interesting,
>
> http://www.coreboot.org/EHCI_Debug_Port
>
> It is already used in the linux kernel and there is also a linux
> module that acts as a debug monitor/device on the other side.
>
> https://www.kernel.org/doc/Documentation/x86/earlyprintk.txt and
> http://www.coreboot.org/EHCI_Gadget_Debug
Sounds interesting and useful. It looks as though the actual hardware is relatively expensive (~US$100 for a NET20DC). If anyone wants to donate me one of these (or find me a cheaper supply), I'll happily implement this feature.
Michael
------------------------------
Message: 3
Date: Thu, 7 Nov 2013 10:07:28 -0800
From: Larry Brigman <larry.brigman at gmail.com>
Cc: ipxe-devel at lists.ipxe.org
Subject: Re: [ipxe-devel] undi interface info
Message-ID:
<CAJEEjpbBcN2vaQOT=jd30ry-OvYtgy6dpmYY_Jhj3s6xx3fySw at mail.gmail.com>
Content-Type: text/plain; charset="iso-8859-1"
I tested it without the embedded script and it works too.
It isn't readily obvious from the docs that chain loading undionly.kpxe causes the above actions.
On Wed, Nov 6, 2013 at 4:49 PM, Larry Brigman <larry.brigman at gmail.com>wrote:
> Replaced ipxe.pxe with undionly.kpxe and the same script and it works
> without the extra cycling through all the interfaces.
>
> Thanks.
>
>
>
> On Wed, Nov 6, 2013 at 4:01 PM, Larry Brigman <larry.brigman at gmail.com>wrote:
>
>> Does that need to be undionly.pxe file or some form of keep pxe?
>> I ask because I embedded the following script with ipxe.pxe
>> ---------
>> #!ipxe
>> autoboot netX ||
>> shell
>> ----------------
>> and get the following during the chaining process:
>> -----------------------------
>> CLIENT MAC ADDR: 00 30 48 BB AB 4C GUID: 534D4349 0002 BB48 3000
>> BB4830004CAB CLIENT IP: 134.242.160.142 MASK: 255.255.252.0 DHCP
>> IP:
>> 134.242.160.248
>> GATEWAY IP:
>> 134.242.160.254
>> PXE->EB: !PXE at 94CB:0070, entry point at
>> 94CB:0106
>> UNDI code segment 94CB:44B0, data segment 8E92:6390
>> (570-613kB)
>> UNDI device is PCI 06:00.0, type
>> DIX+802.3
>> 613kB free base memory after PXE unload iPXE initialising
>> devices...ok
>>
>>
>>
>>
>>
>> iPXE 1.0.0+ (7405) -- Open Source Network Boot Firmware --
>> http://ipxe.org
>> Features: HTTP iSCSI DNS TFTP AoE SRP bzImage ELF MBOOT PXE PXEXT
>> Menu
>> "netX": no such network
>> device
>> iPXE>
>>
>>
>>
>> On Wed, Nov 6, 2013 at 2:23 PM, Brian Rak <brak at gameservers.com> wrote:
>>
>>> Use the netX variables. These refer to whatever interface was it
>>> booted off of.
>>>
>>>
>>> On 11/6/2013 4:46 PM, Larry Brigman wrote:
>>>
>>> During chain loading of ipxe is there some way to determine which
>>> interface ipxe was loaded from?
>>>
>>> Without this info, I will need to embed a script to walk through
>>> the smbios/product and then still its a guess at which interface to
>>> use because of a different number of nic cards loaded.
>>>
>>> Instead, Is there a way to determine motherboard nic's?
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> ipxe-devel mailing
>>> listipxe-devel at lists.ipxe.orghttps://lists.ipxe.org/mailman/listinfo
>>> .cgi/ipxe-devel
>>>
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ipxe.org/pipermail/ipxe-devel/attachments/20131107/9818d4c5/attachment-0001.html>
------------------------------
Message: 4
Date: Thu, 07 Nov 2013 23:51:53 +0000
From: Michael Brown <mbrown at fensystems.co.uk>
To: "Petry Roman, SHS-S IT-TS" <Roman.Petry at shsservices.org>
Cc: ipxe-devel at lists.ipxe.org
Subject: Re: [ipxe-devel] suggestion for improvement - USB
Logging/Debug
Message-ID: <527C2799.9030002 at fensystems.co.uk>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
On 07/11/13 17:21, Petry Roman, SHS-S IT-TS wrote:
> 8-)...
> Nice try to get such a device 8-)..
> if i understand the linux ehci gadget debug module rught, itwill make your normal linux pc to a compatible debug device..
> so just take your pc or raspberry pi and connect it via usb to the normal legacy pc you want to debug, and it should work..
Unfortunately it's not that straightforward. You can't hook two computers (USB hosts) together with just a cable; you need a device in the middle which supports the USB Debug Device extensions. For example, the NET20DC (cost US$95):
http://www.semiconductorstore.com/cart/pc/viewPrd.asp?idproduct=12083
Michael
------------------------------
Message: 5
Date: Fri, 8 Nov 2013 11:20:25 +0100
From: "Petry Roman, SHS-S IT-TS" <Roman.Petry at shsservices.org>
To: "Michael Brown" <mbrown at fensystems.co.uk>
Cc: ipxe-devel at lists.ipxe.org
Subject: Re: [ipxe-devel] suggestion for improvement - USB
Logging/Debug
Message-ID:
<116C0EBA85E1C64DA9891D284A992E8B854591 at win84.dh-com.continental.steel>
Content-Type: text/plain; charset="iso-8859-1"
>> Unfortunately it's not that straightforward. You can't hook two computers (USB hosts) together with just a cable; you need a device in the middle which supports the USB Debug Device extensions. For example, the NET20DC (cost US$95):
>> http://www.semiconductorstore.com/cart/pc/viewPrd.asp?idproduct=12083
>> Michael
Hello Michael,
well, what should i say.. you are absolutely right..t?s better to read the text slow 8-) to understand it right on the corebootpage i?m willing to support / donate some euro/dollar via paypal if you want to rise die 90 Dollar fort he net2dc deivce, or 2 of those device to build an own version of such a debug system:
http://www.coreboot.org/DIY_EHCI_debug_dongle (http://www.ebay.de/itm/EZ-USB-FX2LP-CY7C68013A-USB-Development-Core-Board-Module-Logic-Analyzer-EEPROM-/400470811401?pt=LH_DefaultDomain_77&hash=item5d3deba309 ) 10 Euro per Device + some soldering..
i think the NET20DC is not produced anymore..
Someone else ?
What would you prefer ?
Bye
roman
------------------------------
_______________________________________________
ipxe-devel mailing list
ipxe-devel at lists.ipxe.org
https://lists.ipxe.org/mailman/listinfo.cgi/ipxe-devel
End of ipxe-devel Digest, Vol 42, Issue 9
*****************************************
More information about the ipxe-devel
mailing list