linux-kernel - Re: [PATCH] uio_hv_generic: Enable support for slower vmbus device channels

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]

Message-ID: <20221013041515.GA12822@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net>
Date:   Wed, 12 Oct 2022 21:15:15 -0700
From:   Saurabh Singh Sengar <ssengar@...ux.microsoft.com>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     ssengar@...rosoft.com, kys@...rosoft.com, haiyangz@...rosoft.com,
        sthemmin@...rosoft.com, wei.liu@...nel.org, decui@...rosoft.com,
        longli@...rosoft.com, linux-hyperv@...r.kernel.org,
        linux-kernel@...r.kernel.org, mikelley@...rosoft.com
Subject: Re: [PATCH] uio_hv_generic: Enable support for slower vmbus device
 channels

On Wed, Oct 12, 2022 at 05:54:07PM +0200, Greg KH wrote:
> On Wed, Oct 12, 2022 at 05:52:05PM +0200, Greg KH wrote:
> > On Wed, Oct 12, 2022 at 04:56:46AM -0700, Saurabh Sengar wrote:
> > > This patch enables support for slower vmbus channels, which consist
> > > of following 3 changes :
> > > 	1. Support for hypercalls
> > > 	2. Module params for recv/send buffer sizes
> > > 	3. Module param for custom ring buffer sizes
> > 
> > Even if this all was ok, you are doing 3 things all in one change,
> > that's not allowed at all, you all know this.
> > 
> > Anyway, no new module parameters, this is not the 1990's, we have much
> > better ways to do this properly (hint, module parameters modify code,
> > you want to modify the options of a specific device.)
> 
> Also, you give no good reason for why this is needed at all, nor how
> anyone would use these options and why they would need to.
> 
> The kernel should "just work" and not require manual intervention by a
> user.  Dynamically fix this based on the device, do NOT force a user to
> have to attempt to "tune" anything, that will never work properly over
> time, AND you are being lazy and forcing each individual user to do the
> work, making more effort needed overall than just doing it properly in
> the kernel.

Let me find a method if we can avoid using module parameters, this may result
in hardcoding values in the vmbus driver code for various devices, giving less
flexibilty to user. Meanwhile I figure out this, we can go ahead with
"support for hypercalls", I will send a new patch for it.

Regards,
Saurabh

> 
> thanks,
> 
> greg k-h