Linux 2.6.0
Quoted 5 lines · Andrew Morton
It's actually rather short because I started late. See below.
There are also the "must-fix" and "should-fix" lists of items which we have identified as still on the 2.6 todo list. These are at
Ok, a few updates on my side :
Quoted 6 lines
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/must-fix/must-fix-7.txt and
drivers/net/irda/
~~~~~~~~~~~~~~~~~o dongle drivers need to be converted to sir-dev
In (slow) progress ; tekram-sir patches on my web page.
Quoted 1 line
o irport need to be converted to sir-kthread
Not done.
Quoted 1 line
o new drivers (irtty-sir/smsc-ircc2/donauboe) need more testing
In (slow) progress ; Patch for irtty-sir on my web page.
Quoted 3 lines
o rmk: Refuse IrDA initialisation if sizeof(structures) is incorrect (I'm not sure if we still need this; I think gcc 2.95.3 on ARM shows this problem though.)
Included in 2.6.0-test1 (file net/irda/irlap.c, line 80) Additions :
net/irda/*
~~~~~~~~~~
o F-timer expiry fix ; patch on my web page
o IrCOMM disconnect race condition Oops ; todo <Those are short term items and under control, so don't need
to go into your todo list>
Quoted 6 lines
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/must-fix/should-fix-7.txt
drivers/net/wireless/
~~~~~~~~~~~~~~~~~~~~~o get latest orinoco changes from David.
2.6.0 has the latest stable version (as version 0.14 is still
beta). David may want to comment on his plans.
Quoted 2 lines
o get the latest
airo.cfixes from CVS. This will hopefully fix problems people have reported on the LKML.
Javier has done a great job, cf. fixes in 2.6.0-final.
Quoted 1 line
o get HostAP driver in the kernel.
Todo.
Quoted 1 line
o get more wireless drivers into the kernel.
Todo. Atmel Pcmcia done. The release of 2.6.0 will help get
maintainer motivated ;-)
Quoted 4 lines
o The last two drivers mentioned above are held up by firmware issues (see flamewar on LKML a few days ago). So maybe fixing those firmware issues should be a requirement for 2.6.X, because we can expect more wireless devices to need firmware upload at startup coming to market.
Done. The Atmel wireless driver in the kernel is using the new
firmware loading infrastructure (as well as other drivers), and distro are picking up firmare.agent hotplug script.
Quoted 7 lines
drivers/net/
~~~~~~~~~~~~o davej: Either Wireless network drivers or PCMCIA broke somewhen. A configuration that worked fine under 2.4 doesn't receive any packets. Need to look into this more to make sure I don't have any misconfiguration that just 'happened to work' under 2.4
Please report those bugs to the driver authors with precise
details of setup and issue. Such useless bug report should not be in a todo list, as it's impossible to act on it.
I'll probably send you stuff for 2.6.1 when I'll be back from
the hollyday break.
Have fun... Jean
Quoted 6 lines · Jean Tourrilhes
Quoted 4 lines
drivers/net/irda/
~~~~~~~~~~~~~~~~~o dongle drivers need to be converted to sir-dev
In (slow) progress ; tekram-sir patches on my web page.
I have converted dongle drivers on disk here. So far I was holding back basically due to:
* 2.6.0 freeze
* neither hardware nor testers to do even basic validation
* wait for tx-lock and rawmode fixes to get applied first so we would hopefully get some more feedback and trust in the infrastructure.
If you would like to accept it now I'd sent patches this weekend.
Quoted 3 lines
Quoted 1 line
o new drivers (irtty-sir/smsc-ircc2/donauboe) need more testing
In (slow) progress ; Patch for irtty-sir on my web page.
Yes, tekram-sir is reportedly working now with those patches.
Quoted 1 line
o IrCOMM disconnect race condition Oops ; todo
After some user provided helpful information off-list I found out it gets triggered when the app doesn't close the ircomm-fd before exiting. Just do "cat /dev/ircomm0" and kill -9 the cat to see.
Below my current experimental patch. It works for me and I've got an user confirmation meanwhile. It's definedly an improvement and I do think it is as safe as we can get without reworking the whole ircomm-locking...
Martin
----------------------
net/irda/ircomm/ircomm_tty.c +2 -2
@@ -982,6 +982,8 @@ static void ircomm_tty_shutdown(struct iif (!(self->flags &ASYNC_INITIALIZED ))return ; +ircomm_tty_detach_cable (self);/* first it does is deleting wd-timer! */ +spin_lock_irqsave (&self->spinlock, flags);del_timer (&self->watchdog_timer);
@@ -998,8 +1000,6 @@ static void ircomm_tty_shutdown(struct i self->tx_skb =NULL ; } -ircomm_tty_detach_cable (self); -if (self->ircomm) {ircomm_close (self->ircomm); self->ircomm =NULL ;
Quoted 7 lines · Martin Diehl
I have converted dongle drivers on disk here. So far I was holding back basically due to:
* 2.6.0 freeze
* neither hardware nor testers to do even basic validation
* wait for tx-lock and rawmode fixes to get applied first so we would hopefully get some more feedback and trust in the infrastructure.
Yep. Hopefully this will change as more user migrate to 2.6.X.
Quoted 1 line
If you would like to accept it now I'd sent patches this weekend.
I'm flying tommorow, and when I'm away, I never read my e-mail.
Quoted 9 lines
Quoted 1 line
o IrCOMM disconnect race condition Oops ; todo
After some user provided helpful information off-list I found out it gets triggered when the app doesn't close the ircomm-fd before exiting. Just do "cat
/dev/ircomm0" and kill -9 the cat to see.Below my current experimental patch. It works for me and I've got an user confirmation meanwhile. It's definedly an improvement and I do think it is as safe as we can get without reworking the whole ircomm-locking...
Yes, that is more or less what I had suggested. Note that I'm unhappy about (self->flags & ASYNC_INITIALIZED),
either it goes under spinlock or use test_bit().
Quoted 1 line
Martin
Jean