X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C880CC.15C43B18@onstor-exch02.onstor.net>; Fri, 7 Mar 2008 19:25:40 -0800
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Content-class: urn:content-classes:message
Subject: RE: Review request : Defect  TED00020613
Date: Fri, 7 Mar 2008 19:25:39 -0800
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E08C0F53E@onstor-exch02.onstor.net>
In-Reply-To: <20080307184340.303392b8@ripper.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Review request : Defect  TED00020613
Thread-Index: AciAxjmcjUnuoGVcSBq4Jx7zjw2H/AAAIJtA
References: <BB375AF679D4A34E9CA8DFA650E2B04E08C0F3DC@onstor-exch02.onstor.net> <20080307184340.303392b8@ripper.onstor.net>
From: "Kumar Vakacharla (HCL)" <kumarv@onstor.com>
To: "Andy Sharp" <andy.sharp@onstor.com>
Cc: "Tim Gardner" <tim.gardner@onstor.com>

Andy,

Please find below my responses for your comments.

Ssc-support/support.sh:

* Looks like someone already beat you to the punch on this one, but
* perhaps your change is better.  Have you verified that "echo -e" works
* as expected on OpenBSD as well as Linux?

Kumar> Yes, there was a check-in (28241) happened in the afternoon which
also introduced the similar fix. As you said I will keep my change. =20
I have tested "echo -e" in OpenBSD and Linux and its working expected.


* Line 1964, you changed a variable name from TEMPFILE2 to
TEMPFILE2_BASE
* which I don't think you want to do.

Kumar> Corrected. Actually this was not my change. It was TEMPFILE2_BASE
earlier itself.  Recent check-in has corrected it to TEMPFILE2.


I was not aware of the process of creating changelists and send them for
review.=20

I will follow your suggestions from now onwards. Thank You


-Kumar.

-----Original Message-----
From: Andy Sharp=20
Sent: Friday, March 07, 2008 6:44 PM
To: Kumar Vakacharla (HCL)
Cc: Tim Gardner
Subject: Re: Review request : Defect TED00020613

On Fri, 7 Mar 2008 14:35:08 -0800 "Kumar Vakacharla (HCL)"
<kumarv@onstor.com> wrote:

> Andy,=20
>=20
> =20
>=20
>             Can you please review this fix.=20
>=20
>            =20
>=20
> Thanks,
>=20
> Kumar.
>=20
> =20
>=20
> ________________________________
>=20
> From: Tim Gardner=20
> Sent: Friday, March 07, 2008 2:31 PM
> To: Kumar Vakacharla (HCL)
> Subject: RE: Review request : Defect TED00020613
>=20
> =20
>=20
> Please have Andy Sharp do the code review.
>=20
> Thanks for your effort on this.
>=20
> 	=20
>=20
> =09
> ________________________________
>=20
>=20
> 	From: Kumar Vakacharla (HCL)=20
> 	Sent: Friday, March 07, 2008 2:10 PM
> 	To: Tim Gardner
> 	Subject: Review request : Defect TED00020613
>=20
> 	Tim,
>=20
> 	=20
>=20
> 	I have done a fix for the defect TED00020613 and it needs to
> be reviewed. =20
>=20
> 	Please let me know whom can I send it for review.
>=20
> 	=20
>=20
> 	Here are the details about the fix:
>=20
> 	=20
>=20
> 	=20
>=20
> 	Defect Description:=20
>=20
> 	=20
>=20
> 	can't create two system get alls without renaming the first
>=20
> 	=20
>=20
> 	=20
>=20
> 	Cause:=20
>=20
> 	=20
>=20
> 	This is happening only in Cougar machines.  Generally <.N> is
> appended if there are more than one directories output by=20
>=20
> 	support.sh in one day. But this behavior is failing in Linux
> because of the following two reasons.=20
>=20
> 	=20
>=20
> 	1.       In Linux, echo command doesn't interpret backslash
> escapes unless "-e" option is provided.  In the support.sh while
> creating new directory (in function setup_new_target_dir)=20
>=20
> 	      "echo" has been used (without "-e" option) to pass the
> arguments to the binary filesh.   This is working fine in bobcat and
> cheetah because this feature is enabled by default (irrespective of
> "-e" option) in=20
>=20
> 	      shell "sh" used by openbsd.=20
>=20
> 	=20
>=20
> 	2.       Having resolved the above issue, there is one more
> issue in Linux which is related to the alignment of the buffer
> containing directory entry records. =20
>=20
> 	In function cmd_readdir (in sm-file/filesh.c) we use a global
> character buffer readBuf which is used to hold the directory entry
> records. This buffer has been populated with the record data=20
>=20
> 	using memcpy in the function file_req_read_dir in
> sm-file/file-server.c.=20
>=20
> 	=20
>=20
> 	Each directory record will be aligned to 8
> bytes(sizeof(uint64)) internally. We use this rule to traverse to the
> next directory record in the buffer.=20
>=20
> 	But in Linux I found that starting address of readBuf is not 8
> byte aligned but only 4 byte aligned and this has triggered the
> problem.
>=20
>=20
> 	Because of this it is traversing to an offset which may not be
> the beginning of the next record and hence reading the directory
> entries fail eventually.=20
>=20
> 	=20
>=20
> 	Files Modified:=20
>=20
> 	=20
>=20
> =09
> /homes/kumarv/work/dev/nfx-tree/code/ssc-support/support.sh

Looks like someone already beat you to the punch on this one, but
perhaps your change is better.  Have you verified that "echo -e" works
as expected on OpenBSD as well as Linux?

Line 1964, you changed a variable name from TEMPFILE2 to TEMPFILE2_BASE
which I don't think you want to do.

 =09
> /homes/kumarv/work/dev/nfx-tree/code/sm-file/filesh.c=20

looks ok


> 	=20
>=20
> 	=20
>=20
> 	P4CLIENT=3Dkumarv-DEV
>=20
> 	Location: /homes/kumarv/work/dev           =20
>=20
> 	=20
>=20
> 	Please let me know if you need any clarifications.

With just two files, it wasn't that big a deal, but in the future,
capture your changes into a changelist and just send me the changelist
number.  My usual convention is to also send the output of the command
"p4 describe <cl#>" so the reviewer can see the list of files and the
checkin comment that will accompany the change.  If you are unfamiliar
how to do this, see the perforce command "p4 change" or do it using the
GUI.  In the GUI, when you select some files and click on [submit] it
gives you the option to save it to a changelist number.

Cheers,

a


