Syslinux - Mar 2014 - Very slow download with pxelinux > 4.07 on specific hardware

On Mon, Mar 10, 2014 at 7:15 PM, Gene Cumm <gene.cumm at gmail.com>
wrote:
> On Mon, Mar 10, 2014 at 6:52 PM, Schlomo Schapiro <schlomo at
schapiro.org> wrote:
>> Hi,
>>
>> On 10 March 2014 18:13, Eric PEYREMORTE <eric.peyremorte at
iut-valence.fr>wrote:
>>
>>> Ok i give up. I will stick with the old version version (4.07)
which works
>>> well with dell 380 and 790.
>>>
>>> I tried to understand what happened between 4.07 and 5 but i
don't have
>>> the skill to understand low level c.
>>>
>>> Seems that the memory management changed a lot, maybe the culprit.
But as
>>> many other things changed this is not obvious...
>>>
>>
>> sad to say, we have exactly the same problem. With HTTP downloads it is
>> much worse, we could not find a stable and reliable version after 4.x.
>> Sadly nobody wanted to fix our syslinux problem, even as paid
development.
>>
>> I very much hope that the syslinux project will overcome these
obstacles.
>> It would be sad to see all the new development stay unused because if
this.
>
> It's also a balance of time.  While working on 4.10-pre*/5.10-pre*, I
> found that some hardware misreports its behavior.  "Sure, Interrupts
> work" but they don't is but one that I worked around on specific
> hardware.
>
> --
> -Gene
Simple test (and boot failure is an option as long as the core's
messages are intact).  Could both of you try fetching
Syslinux-4.10-pre22 (zip versus tar is your choice), extract out
core/pxelinux.0 and boot with it then record the 4-digit value for
flags (in the "UNDI: baseio" line) and the first 3 octets (6 hex
digits including 0s) of the MAC address and take a reasonable screen
shot (I find a cell phone against a solid object to be pretty decent
when angled to reduce glare and distanced for proper focus) but don't
bother sending the picture yet.  It's possible that the workaround I
utilized can either be having a negative impact or needs to be
extended to more hardware.

Attached you'll find an image with the two pieces highlighted.

--
-Gene
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 140310-1950-sl410p22-2.png
Type: image/png
Size: 12009 bytes
Desc: not available
URL:
<http://www.zytor.com/pipermail/syslinux/attachments/20140310/92f6856c/attachment.png>

On Mon, Mar 10, 2014 at 7:55 PM, Gene Cumm <gene.cumm at gmail.com>
wrote:
> On Mon, Mar 10, 2014 at 7:15 PM, Gene Cumm <gene.cumm at gmail.com>
wrote:
>> On Mon, Mar 10, 2014 at 6:52 PM, Schlomo Schapiro <schlomo at
schapiro.org> wrote:
>>> Hi,
>>>
>>> On 10 March 2014 18:13, Eric PEYREMORTE <eric.peyremorte at
iut-valence.fr>wrote:
>>>
>>>> Ok i give up. I will stick with the old version version (4.07)
which works
>>>> well with dell 380 and 790.
>>>>
>>>> I tried to understand what happened between 4.07 and 5 but i
don't have
>>>> the skill to understand low level c.
>>>>
>>>> Seems that the memory management changed a lot, maybe the
culprit. But as
>>>> many other things changed this is not obvious...
>>>>
>>>
>>> sad to say, we have exactly the same problem. With HTTP downloads
it is
>>> much worse, we could not find a stable and reliable version after
4.x.
>>> Sadly nobody wanted to fix our syslinux problem, even as paid
development.
>>>
>>> I very much hope that the syslinux project will overcome these
obstacles.
>>> It would be sad to see all the new development stay unused because
if this.
>>
>> It's also a balance of time.  While working on 4.10-pre*/5.10-pre*,
I
>> found that some hardware misreports its behavior.  "Sure,
Interrupts
>> work" but they don't is but one that I worked around on
specific
>> hardware.
>>
>> --
>> -Gene
>
> Simple test (and boot failure is an option as long as the core's
> messages are intact).  Could both of you try fetching
> Syslinux-4.10-pre22 (zip versus tar is your choice), extract out
> core/pxelinux.0 and boot with it then record the 4-digit value for
> flags (in the "UNDI: baseio" line) and the first 3 octets (6 hex
> digits including 0s) of the MAC address and take a reasonable screen
> shot (I find a cell phone against a solid object to be pretty decent
> when angled to reduce glare and distanced for proper focus) but don't
> bother sending the picture yet.  It's possible that the workaround I
> utilized can either be having a negative impact or needs to be
> extended to more hardware.
>
> Attached you'll find an image with the two pieces highlighted.
>
> --
> -Gene
https://www.kernel.org/pub/linux/utils/boot/syslinux/Testing/4.10/

Le 11/03/2014 00:56, Gene Cumm a ?crit :
> On Mon, Mar 10, 2014 at 7:55 PM, Gene Cumm <gene.cumm at gmail.com>
wrote:
>> On Mon, Mar 10, 2014 at 7:15 PM, Gene Cumm <gene.cumm at
gmail.com> wrote:
>>> On Mon, Mar 10, 2014 at 6:52 PM, Schlomo Schapiro <schlomo at
schapiro.org> wrote:
>>>> Hi,
>>>>
>>>> On 10 March 2014 18:13, Eric PEYREMORTE <eric.peyremorte at
iut-valence.fr>wrote:
>>>>
>>>>> Ok i give up. I will stick with the old version version
(4.07) which works
>>>>> well with dell 380 and 790.
>>>>>
>>>>> I tried to understand what happened between 4.07 and 5 but
i don't have
>>>>> the skill to understand low level c.
>>>>>
>>>>> Seems that the memory management changed a lot, maybe the
culprit. But as
>>>>> many other things changed this is not obvious...
>>>>>
>>>> sad to say, we have exactly the same problem. With HTTP
downloads it is
>>>> much worse, we could not find a stable and reliable version
after 4.x.
>>>> Sadly nobody wanted to fix our syslinux problem, even as paid
development.
>>>>
>>>> I very much hope that the syslinux project will overcome these
obstacles.
>>>> It would be sad to see all the new development stay unused
because if this.
>>> It's also a balance of time.  While working on
4.10-pre*/5.10-pre*, I
>>> found that some hardware misreports its behavior.  "Sure,
Interrupts
>>> work" but they don't is but one that I worked around on
specific
>>> hardware.
>>>
>>> --
>>> -Gene
>> Simple test (and boot failure is an option as long as the core's
>> messages are intact).  Could both of you try fetching
>> Syslinux-4.10-pre22 (zip versus tar is your choice), extract out
>> core/pxelinux.0 and boot with it then record the 4-digit value for
>> flags (in the "UNDI: baseio" line) and the first 3 octets (6
hex
>> digits including 0s) of the MAC address and take a reasonable screen
>> shot (I find a cell phone against a solid object to be pretty decent
>> when angled to reduce glare and distanced for proper focus) but
don't
>> bother sending the picture yet.  It's possible that the workaround
I
>> utilized can either be having a negative impact or needs to be
>> extended to more hardware.
>>
>> Attached you'll find an image with the two pieces highlighted.
>>
>> --
>> -Gene
First, 4.10-pre22 works as well as 4.07 on my dell optiplex 380. It is 
slow as hell from v5 to 6.03.
I also tried to put a 3com NIC inside, and the result was the same. But 
when i put the nic in an optiplex
740 it boots fine.

CLIENT MAC ADDR: B8 AC 6F
UNDI : baseio 0000 int 11 MTU 1500 type 1 "DIX+802.3" flags 0x1c1b

Ps : sorry for the capture quality i had to make a movie and extract a 
picture cause it was booting too fast.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Capture.JPG
Type: image/jpeg
Size: 138779 bytes
Desc: not available
URL:
<http://www.zytor.com/pipermail/syslinux/attachments/20140311/d427c3e3/attachment-0001.jpe>