Re: [Acx100-devel] [PATCH 00/30] on v20120506

[Oliver W. <oli...@ol...>]

Hi Jim,

On 05/08/2012 10:09 PM, Jim Cromie wrote:
> Another workflow you might consider:
> make a branch at the same point the patchset starts,
> git am 00*  into that branch,
> git rebase master branch
> 
> with this, you get similar conflict markers right in the troubled
> code, and can just delete
> the bad chunk.  It may be fewer steps (I dunno about patch --merge, so
> cant compare them)
> you can also rebase --abort, and redo the rebase on any commit 1/2 way
> between start and master - it doesnt eliminate the work, but may
> reduce the complexity of some of the changes.

Ok, thanks for the tip :)!

>> Question regarding the struct acx_device refactoring, just to be sure:
>>
>> Does this still allow concurrent operation of multiple and mixed
>> device-types (i.e. several pci, usb, mem devices on the on the same
>> machine, each one with it's own adev, running at the same time) ?
>>
> 
> I dont think it changes anything in that regard, but I dont Know.
> I recall wondering (or asking, not sure if I did) about driving
> multiple devices,

Ok, good ;). I'll just was about trying, and it seems to probe ok so far.
Will still test in more detail then.
---
> lapix:~/Test# iwconfig 
[...]
> wlan4     IEEE 802.11b  ESSID:off/any  
>           Mode:Managed  Access Point: Not-Associated   Tx-Power=0 dBm   
>           Retry  long limit:7   RTS thr:off   Fragment thr:off
>           Encryption key:off
>           Power Management:on
>           
> wlan3     IEEE 802.11bg  ESSID:off/any  
>           Mode:Managed  Access Point: Not-Associated   Tx-Power=0 dBm   
>           Retry  long limit:7   RTS thr:off   Fragment thr:off
>           Encryption key:off
>           Power Management:on
---

> For this, you end up with multiple acx_device structs right ?
> (otherwize youd not have a wep keyset for each)

Indeed, basically it's just the normal allocation of a new device
private struct (acx_device struct) upon probe of each device.

> I changed the composition/aggregation of acx_device_t, but all the pieces
> are there.  Similar to moving cfg_opts fields into a separate struct.
> 
> Can you comment on how multi-device scenario works ?
> My attempt to get a working USB-only compile might conflict with
> (or be informed by) this understanding.

>From the practical point of view (but maybe I didn't understood the
question well, just tell in case of), e.g. for the case above: it's just
a acx111 DWL-G650+ pci card and a DWL-120+ connected on the same laptop.
That results in two wifi phys and interfaces and can be used
independently from each other.

Now I agree, that's not a very typical use-case, but e.g. one could
think about an router setup with multiple wifi devices to do something
like similar WDS (with multiple radios) ... .

Cheers, Oliver ;)