cvs commit: src/sys/dev/ata ata-queue.c

[Søren Schmidt]

On 28/06/2005, at 21:20, John Baldwin wrote:

> On Tuesday 28 June 2005 02:24 pm, Søren Schmidt wrote:
>
>> On 28/06/2005, at 19:10, John Baldwin wrote:
>>
>>> On Tuesday 28 June 2005 11:30 am, Søren Schmidt wrote:
>>>
>>>> On 28/06/2005, at 15:51, John Baldwin wrote:
>>>>
>>>>> On Tuesday 28 June 2005 05:06 am, SXren Schmidt wrote:
>>>>>
>>>>>> sos         2005-06-28 09:06:52 UTC
>>>>>>
>>>>>>   FreeBSD src repository
>>>>>>
>>>>>>   Modified files:
>>>>>>     sys/dev/ata          ata-queue.c
>>>>>>   Log:
>>>>>>   Zero donecount on auto request sense.
>>>>>>
>>>>>>   PR:             81450
>>>>>>   Approved by:    re@ (scottl)
>>>>>>
>>>>>
>>>>> Are you going to commit this to 5.x now as well?  FWIW, the patch
>>>>> in question
>>>>> was straight from the bug report as well.
>>>>>
>>>>
>>>> Well, I thought that the plan was to have 6.0 be the solution to  
>>>> 5.x
>>>> problems ;)
>>>>
>>>> Anyhow if/when I'll commit anything to 5.x, it will be the ATA  
>>>> driver
>>>> from 6.0/current.
>>>> The problem being that the ABI for atacontrol etc has changed so it
>>>> kindof breaks the charter of -stable IMHO.
>>>> Other than that I have the bits sitting here on my lone -stable box
>>>> just waiting for a push on the big red commit key :)
>>>> .
>>>> - Søren
>>>>
>>>
>>> Well is it ok if I merge just this change to 5.x then?
>>>
>>
>> As I've stated earlier I don't support what's been put into 5.x to
>> "fix" bugs.
>> ATA mkIII is the fix for the 5.x problems/bugs from this end, so you
>> can do exactly what you want on the ATA code in 5.x as I don't really
>> care :)
>>
>
> I'll be sure to remember that helpful attitude the next time you  
> have an issue
> with one of your production machines that I could help with.  Also,  
> given
> that you committed the exact patch from the PR to HEAD and then  
> claimed when
> you closed the PR prematurely that it was "solved (differently) in - 
> current"
> that was very rude to the submitter who took time to find a bug in  
> *your*
> code and submit a working patch to fix it.

Just to give the balance here I have mailed with the submitter and  
excused that it wasn't in -current but only in my local tree, shit  
happens you know..

> I've also offered numerous times to do the actual commit of the fix  
> to 5.x if
> you would give it a quick glance over but you always responded to  
> both me and
> the submitter by saying that the bug was already fixed in ata mkIII  
> and
> wouldn't comment on the validity of the patch other than to say  
> that the bug
> was fixed differently in a different file in current.  Given that  
> you just
> now committed the exact patch to HEAD, it would seem that, in fact,  
> ata mkIII
> did _not_ contain the correct fix as you had previously stated, and  
> I guess
> the fact that you committed it to HEAD finally gives me some actual  
> feedback
> on my requests for you to give it a quick review so the fix could  
> be put in
> 5.x (since I was under the impression from your earlier e-mails  
> that this
> issue was present on 5.x only).

Right, and all the mess is because it was decided to hack around in  
ATA in 5.x whilst it was well known that the mkIII work was close to  
being done. Mind you mkIII is all about fixing bugs in ATA, and  
finetuning the infrastructure for future work.
That made me abandon ATA as is in 5.x, as I only have 24 hours a day  
and a real life to take care off (I know that is not the common case  
around here), and I simply don't have the time to sort out what hss  
been randomly thrown into the pot, sorry. If the project cannot live  
with that, I'll hand in my maintainer and commit bit any day, just  
say when, I'm tired so tired already..

Now, I do have a 5.4 version of ATA from current, but its unclear to  
me if it is wanted/allowed into 5.x, and until that eventual commit  
happens I'm out of the loop on ATA in 5.x which I also stated  
repeatedly here on the lists.

That all said I seriously think we all need a vacation and several  
nights of good sleep :)

- Søren