Search Linux Wireless

Re: [PATCH] iwlwifi: get rid of IWL_{GET,SET}_BITS crap

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Oct 8, 2008 at 10:02 AM, Johannes Berg
<johannes@xxxxxxxxxxxxxxxx> wrote:
> On Tue, 2008-10-07 at 22:46 +0200, Tomas Winkler wrote:
>> >> > bytes (with nibbles): | . | . | . | . | . | . || . | . | . | . | . | . |
>> >> >                      | addr 1          |len1 || addr 2          |len2 |
>> >> > your forced layout:   |               |                |               |
>> >> > my layout:            |               |       ||               |       |
>> >> >
>> >> > Note how this actually matches the border between the border between the
>> >> > two descriptors, the || border.
>> >>
>> >> Your layout put addr2 (which is 32 bit) sits on 16 bit boundary
>> >
>> > So? Yours makes it need two words, which is even less efficient to
>> > access. And since the struct is packed, on those architectures that
>> > don't do unaligned accesses efficiently won't need any put_unaligned
>> > either.
>> >
>> > Also, keep in mind that address 2 is _never_ used at all anyway
>
>> This is not correct we always use first 2 pointers. First for tx command
>> second for the actual packet.
>
> Ok so you use two.

My only point is just to make sure that you understand that __never__
is not correct

 Writing three words in total. I strongly suggest that
> you have WAY more trouble in iwlwifi than an imagined performance issue
> coming from a corrected and understandable struct layout.

You are rally trying breaking into open doors Currently I'm more
concern with correctness then performance so I wanted to rise
hopefully all issues.
I'm testing your layout it's work so far in my home setup. I'm on
holidays till EOW so I will be able to give it some more stress in Lab
only next week.

>
>> > because
>> > we don't and cannot do fragmented frames due to the lack of IP
>> > checksumming in hardware, as I and davem have tried to explain to you
>> > multiple times already.
>>
>> This is bad side of current Linux implementation
>
> Look, we've tried to explain it to you so many times that it simply is
> not possible to enable fragment collection without IP checksumming in
> hardware that I don't know what to tell you. This may sound offensive,
> but are you really too stupid to understand it?

Okay not fair from my side I just couldn't resist ... I just really
like the subject :)
Tomas
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[Index of Archives]     [Linux Host AP]     [ATH6KL]     [Linux Bluetooth]     [Linux Netdev]     [Kernel Newbies]     [Linux Kernel]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Device Mapper]
  Powered by Linux