Re: [edk2] Using multiple SNP drivers on Shell

David Van Arnem Tue, 05 Apr 2016 10:33:59 -0700

On 04/04/2016 10:30 PM, Andrew Fish wrote:

On Apr 4, 2016, at 9:17 PM, Bhupesh Sharma <bhupesh.sha...@nxp.com> wrote:

-----Original Message-----
From: Carsey, Jaben [mailto:jaben.car...@intel.com]
Sent: Monday, April 04, 2016 11:28 PM
To: Laszlo Ersek; Bhupesh Sharma
Cc: edk2-devel@lists.01.org; David Van Arnem; Carsey, Jaben
Subject: RE: [edk2] Using multiple SNP drivers on Shell

-----Original Message-----
From: edk2-devel [mailto:edk2-devel-boun...@lists.01.org] On Behalf Of
Laszlo Ersek
Sent: Monday, April 04, 2016 10:47 AM
To: Bhupesh Sharma <bhupesh.sha...@nxp.com>
Cc: edk2-devel@lists.01.org <edk2-de...@ml01.01.org>; David Van Arnem
<dvanar...@cmlab.biz>
Subject: Re: [edk2] Using multiple SNP drivers on Shell
Importance: High

On 04/04/16 19:38, Bhupesh Sharma wrote:

From: Laszlo Ersek [mailto:ler...@redhat.com]
Sent: Monday, April 04, 2016 11:06 PM

On 04/04/16 19:24, David Van Arnem wrote:

On 04/04/2016 03:17 AM, Laszlo Ersek wrote:

The most recent version (that I have) of the Shell Spec is 2.2.
That version lists both "ping" and "ping6", and the synopsis for

"ping" is:


   Ping [-n count] [-l size] [-s SourceIp] TargetIp


Interesting, it looks like we each have a different "ping" command:
Bhupesh's requires "-_s [SourceIP]", Laszlo's uses "-s
[SourceIP]", and mine (not sure which commit it was build from, I
built it a while
ago) says to use "-s [interface name]" (this may work with the
Source IP also, but the "help" gave eth0 as an example).


Note: I didn't test "my" ping command, only proved it right ^W^W^W

quoted

the shell spec. :)

But, I suppose that is a moot point, since I should be using an
up-to-date Shell as Laszlo recommended.


The shell source code still uses L"-_s" for the ping (v4) command;
only the ping6 command uses L"-s". Compare:

- ShellPkg/Library/UefiShellNetwork1CommandsLib/Ping.c
- ShellPkg/Library/UefiShellNetwork2CommandsLib/Ping6.c

And L"-_s" in the former dates back to this commit:

commit 68fb05272b45b473fe030ac11dcb79d3f3c0aecd
Author: jcarsey <jcarsey@6f19259b-4bc3-4df7-8a09-765794883524>
Date:   Fri Mar 25 20:47:02 2011 +0000

    Add Network1 profile.

    git-svn-id:
https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11423
6f19259b-4bc3-4df7-8a09-765794883524

The following command:

  git grep 'L"-_' -- ShellPkg

returns a few more hits. I'm thinking the "-_option" pattern is for
non- standard options, and the underscore is supposed to prevent
conflicts with any future standardization.

Maybe in the case of the Ping command, L"-_s" could now be simply
replaced with L"-s".


IMHO, -_s makes no sense. -s is definitely concise and more readable.


Well, the question is whether the behavior of the historical "-_s"
option matches that of the recently standardized "-s" option.

If so, then it's a simple rename.

If the behaviors are different, then "-s" should be implemented
separately, and "-_s" might be preferable to preserve also, as a
non-standard extension.


All non-standard parameters are started with _ per shell specification.

If the behaviors are different it's certainly worth discussing making
sure that needs are met.  The whole goal of making it an official
parameter was to make backwards compatibility guaranteed.


Well, I think if the SNP protocol stack cannot start a PING transmission when
two or more Ethernet interfaces are available for usage, then PING '-_s' is no
longer a non-standard extension.

Presently and in future, we will have multiple Ethernet interfaces available on
a x86 or ARM board and in that case PING -_s is simply a requirement to check
if the connection to a host is alive.

To look at it from the other perspective, a question could come to mind that if 
this a
non-standard (which I read  as requirement not mandated by the specifications) 
requirement,
then the SNP stack should be fixed to handle multiple Ethernet interfaces.


The are both standards based. The -_* is the namespace for non standard shell 
commands so they will never conflict with an update to the shell spec.

I think the concern here might be that ping will "not work" (from Bhupesh's results, I am unable to test on my end) without -_s when there is more than one active interface that uses SNP. In contrast, ping on Linux will choose an interface when there is more than one active and the interface is not specified with a flag in the ping command.

I think the Linux ping behavior could be replicated in the UEFI Shell ping, but I don't know if it's something that would be desired.


David

Thanks,

Andrew Fish

Regards,
Bhupesh
_______________________________________________
edk2-devel mailing list
edk2-devel@lists.01.org
https://lists.01.org/mailman/listinfo/edk2-devel


_______________________________________________
edk2-devel mailing list
edk2-devel@lists.01.org
https://lists.01.org/mailman/listinfo/edk2-devel



--
Thanks,
David Van Arnem

smime.p7s
Description: S/MIME Cryptographic Signature

_______________________________________________
edk2-devel mailing list
edk2-devel@lists.01.org
https://lists.01.org/mailman/listinfo/edk2-devel

Re: [edk2] Using multiple SNP drivers on Shell

Reply via email to