netdev - RE: [PATCH 3/3] vmxnet3: fix typo setting confPA

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 22 Oct 2010 11:10:42 -0700
From:	Shreyas Bhatewara <sbhatewara@...are.com>
To:	Harvey Harrison <harvey.harrison@...il.com>
CC:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"shemminger@...tta.com" <shemminger@...tta.com>
Subject: RE: [PATCH 3/3] vmxnet3: fix typo setting confPA



> -----Original Message-----
> From: Harvey Harrison [mailto:harvey.harrison@...il.com]
> Sent: Friday, October 22, 2010 10:52 AM
> To: Shreyas Bhatewara
> Cc: netdev@...r.kernel.org; shemminger@...tta.com
> Subject: Re: [PATCH 3/3] vmxnet3: fix typo setting confPA
> 
> On Fri, Oct 22, 2010 at 10:46 AM, Shreyas Bhatewara
> <sbhatewara@...are.com> wrote:
> >> From: Harvey Harrison [mailto:harvey.harrison@...il.com]
> >> One question I had when looking into this, why are you doing
> >> byteswapping in the VMXNET_{READ,WRITE}_BAR{0,1}
> >> macros?
> >
> > Because values need to be written in LE orders in registers. For eg.
> when rx_ring.next2fill index is written to RXPROD reg, it should be
> converted from CPU order to LE.
> 
> Yes, but (unless I'm mistaken) writel always writes out in
> little-endian, so there isn't a need to swap the arguments,
> similarly for readl in the other direction.
> 
> You wouldn't notice this on a little-endian machine obviously, has
> this even bee tried on be?

No, the virtualized NIC vmxnet3 will not be available in BE architectures as far as I can tell. Hence the driver can never be exercised on BE machines. The BE support was just put in the driver to get rid of the compile time X86 dependency in config files because developers often used BE machines to test the kernel builds. 

Best regards.
Shreyas


> 
> Harvey