mbox series

[00/22] windows/netuio: add netuio driver for Windows

Message ID 1597360905-74106-1-git-send-email-navasile@linux.microsoft.com (mailing list archive)
Headers show
Series windows/netuio: add netuio driver for Windows | expand

Message

Narcisa Ana Maria Vasile Aug. 13, 2020, 11:21 p.m. UTC
From: Narcisa Vasile <navasile@microsoft.com>

The NetUIO driver for Windows allows the usermode application
to directly access the hardware and allocates the memory that gets mapped in
usermode.

Anand Rawat (1):
  Updated Source and Project files to use Clang toolset

Harini Ramakrishnan (10):
  Windows DPDK libraries and applications have now been updated to the
    latest public release v18.08, of the main DPDK source.
  doc: remove embedded buffer from Windows UIO ioctl
  Windows DPDK libraries and applications have now been updated to the
    latest public release v18.08, of the main DPDK source.
  doc: change the Windows UIO driver's default security descriptor to
    admin only
  doc: remove lower bound on mapped address from Windows UIO driver
  doc: remove embedded buffer from Windows UIO ioctl
  uio: move SDDL string to INF on Windows
  uio: Fix 64 bit BARs mapping
  license: update headers with BSD 3-clause license:
  doc: updating REAME for NetUIO driver

Jason Messer (2):
  Initial commit of UIO driver for Windows
  Added new core libraries for Windows

Narcisa Vasile (8):
  Update .gitignore and create .gitattributes
  uio: Use local time when verifying INF DriverVer
  uio: Remove co-installers section from inf
  uio: Wrap call into try/except block
  uio: Use request handler that guarantees execution in correct context
  uio: Change the device setup class to a custom one
  Enable DMA remapping through INF directive
  Move all files under windows folder

Thomas Monjalon (1):
  init DPDK repository

Comments

Harini Ramakrishnan Aug. 14, 2020, 2:56 p.m. UTC | #1
On Thu, Aug 13, 2020 at 04:21:23PM -0700, Narcisa Ana Maria Vasile wrote:
> From: Narcisa Vasile <navasile@microsoft.com>
> 
> The NetUIO driver for Windows allows the usermode application
> to directly access the hardware and allocates the memory that gets mapped in
> usermode.
> 
> Anand Rawat (1):
>   Updated Source and Project files to use Clang toolset
> 
> Harini Ramakrishnan (10):
>   Windows DPDK libraries and applications have now been updated to the
>     latest public release v18.08, of the main DPDK source.
>   doc: remove embedded buffer from Windows UIO ioctl
>   Windows DPDK libraries and applications have now been updated to the
>     latest public release v18.08, of the main DPDK source.
>   doc: change the Windows UIO driver's default security descriptor to
>     admin only
>   doc: remove lower bound on mapped address from Windows UIO driver
>   doc: remove embedded buffer from Windows UIO ioctl
>   uio: move SDDL string to INF on Windows
>   uio: Fix 64 bit BARs mapping
>   license: update headers with BSD 3-clause license:
>   doc: updating REAME for NetUIO driver
> 
> Jason Messer (2):
>   Initial commit of UIO driver for Windows
>   Added new core libraries for Windows
> 
> Narcisa Vasile (8):
>   Update .gitignore and create .gitattributes
>   uio: Use local time when verifying INF DriverVer
>   uio: Remove co-installers section from inf
>   uio: Wrap call into try/except block
>   uio: Use request handler that guarantees execution in correct context
>   uio: Change the device setup class to a custom one
>   Enable DMA remapping through INF directive
>   Move all files under windows folder
> 
> Thomas Monjalon (1):
>   init DPDK repository
> 
> -- 
> 2.23.0.vfs.1.1.63.g5a5ad7f

Acked-by: Harini Ramakrishnan <haramakr@linux.microsoft.com>
Dmitry Kozlyuk Aug. 14, 2020, 8:01 p.m. UTC | #2
On Thu, 13 Aug 2020 16:21:23 -0700, Narcisa Ana Maria Vasile wrote:
> From: Narcisa Vasile <navasile@microsoft.com>
> 
> The NetUIO driver for Windows allows the usermode application
> to directly access the hardware

> and allocates the memory that gets mapped in usermode.

It doesn't allocate the buffer anymore, does it?

As far as I understand, you're importing history from dpdk-draft-repo and
then changing file layout. Not sure how this is beneficial, especially
since there are bugfixes along the way. Or are there legal reasons?

Please follow the guidelines on commit formatting, in particular:

* Subject must have a topic ("windows/netuio"?), its length is limited.
* Description is mandatory and should explain the changes.
Narcisa Ana Maria Vasile Aug. 14, 2020, 8:26 p.m. UTC | #3
On Fri, Aug 14, 2020 at 11:01:39PM +0300, Dmitry Kozlyuk wrote:
> On Thu, 13 Aug 2020 16:21:23 -0700, Narcisa Ana Maria Vasile wrote:
> > From: Narcisa Vasile <navasile@microsoft.com>
> > 
> > The NetUIO driver for Windows allows the usermode application
> > to directly access the hardware
> 
> > and allocates the memory that gets mapped in usermode.
> 
> It doesn't allocate the buffer anymore, does it?
> 
> As far as I understand, you're importing history from dpdk-draft-repo and
> then changing file layout. Not sure how this is beneficial, especially
> since there are bugfixes along the way. Or are there legal reasons?
> 
Harini, Omar, please advise if we need to preserve history. If not,
I think it's better to just squash all commits and send the netuio code
as one patch. In this way, I can easily make sure that this one patch
has the right signoff, description, subject formatting etc.

Dmitry, would pushing the netuio code as one patch be ok?

> Please follow the guidelines on commit formatting, in particular:
> 
> * Subject must have a topic ("windows/netuio"?), its length is limited.
> * Description is mandatory and should explain the changes.
Omar Cardona Aug. 14, 2020, 8:45 p.m. UTC | #4
Hi Naty,
IMHO this is small and sufficiently scoped where history (individual buildable patches) are not necessary.  We'd prefer to avoid that overhead here.

DmitryK, 
Please let us know if otherwise.



-----Original Message-----
From: Narcisa Ana Maria Vasile <navasile@linux.microsoft.com> 
Sent: Friday, August 14, 2020 1:26 PM
To: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Cc: dev@dpdk.org; thomas@monjalon.net; haramakr@linux.microsoft.com; Omar Cardona <ocardona@microsoft.com>; pallavi.kadam@intel.com; ranjit.menon@intel.com; Dmitry Malloy (MESHCHANINOV) <dmitrym@microsoft.com>; Narcisa Ana Maria Vasile <Narcisa.Vasile@microsoft.com>
Subject: Re: [PATCH 00/22] windows/netuio: add netuio driver for Windows

On Fri, Aug 14, 2020 at 11:01:39PM +0300, Dmitry Kozlyuk wrote:
> On Thu, 13 Aug 2020 16:21:23 -0700, Narcisa Ana Maria Vasile wrote:
> > From: Narcisa Vasile <navasile@microsoft.com>
> > 
> > The NetUIO driver for Windows allows the usermode application to 
> > directly access the hardware
> 
> > and allocates the memory that gets mapped in usermode.
> 
> It doesn't allocate the buffer anymore, does it?
> 
> As far as I understand, you're importing history from dpdk-draft-repo 
> and then changing file layout. Not sure how this is beneficial, 
> especially since there are bugfixes along the way. Or are there legal reasons?
> 
Harini, Omar, please advise if we need to preserve history. If not, I think it's better to just squash all commits and send the netuio code as one patch. In this way, I can easily make sure that this one patch has the right signoff, description, subject formatting etc.

Dmitry, would pushing the netuio code as one patch be ok?

> Please follow the guidelines on commit formatting, in particular:
> 
> * Subject must have a topic ("windows/netuio"?), its length is limited.
> * Description is mandatory and should explain the changes.
Dmitry Kozlyuk Aug. 14, 2020, 8:57 p.m. UTC | #5
On Fri, 14 Aug 2020 13:26:11 -0700, Narcisa Ana Maria Vasile wrote:
> On Fri, Aug 14, 2020 at 11:01:39PM +0300, Dmitry Kozlyuk wrote:
> > On Thu, 13 Aug 2020 16:21:23 -0700, Narcisa Ana Maria Vasile wrote:  
[...]
> > As far as I understand, you're importing history from dpdk-draft-repo and
> > then changing file layout. Not sure how this is beneficial, especially
> > since there are bugfixes along the way. Or are there legal reasons?
> >   
> Harini, Omar, please advise if we need to preserve history. If not,
> I think it's better to just squash all commits and send the netuio code
> as one patch. In this way, I can easily make sure that this one patch
> has the right signoff, description, subject formatting etc.
> 
> Dmitry, would pushing the netuio code as one patch be ok?

Yes, I think so.
Ranjit Menon Aug. 14, 2020, 10:51 p.m. UTC | #6
On 8/14/2020 1:57 PM, Dmitry Kozlyuk wrote:
> On Fri, 14 Aug 2020 13:26:11 -0700, Narcisa Ana Maria Vasile wrote:
>> On Fri, Aug 14, 2020 at 11:01:39PM +0300, Dmitry Kozlyuk wrote:
>>> On Thu, 13 Aug 2020 16:21:23 -0700, Narcisa Ana Maria Vasile wrote:
> [...]
>>> As far as I understand, you're importing history from dpdk-draft-repo and
>>> then changing file layout. Not sure how this is beneficial, especially
>>> since there are bugfixes along the way. Or are there legal reasons?
>>>    
>> Harini, Omar, please advise if we need to preserve history. If not,
>> I think it's better to just squash all commits and send the netuio code
>> as one patch. In this way, I can easily make sure that this one patch
>> has the right signoff, description, subject formatting etc.
>>
>> Dmitry, would pushing the netuio code as one patch be ok?
> Yes, I think so.

I agree too. This code is going to be committed into a new repo (kmods), 
so there is no real benefit in maintaining prior history.

ranjit m.