From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jim Nelson Date: Mon, 20 Dec 2004 21:35:19 +0000 Subject: Re: [KJ] Re: riscom8: replace cli()/sti() Message-Id: <41C74597.5000109@verizon.net> MIME-Version: 1 Content-Type: multipart/mixed; boundary="===============7401475128657049==" List-Id: To: kernel-janitors@vger.kernel.org --===============7401475128657049== Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Domen Puncer wrote: > On 15/12/04 23:45 -0600, james4765@verizon.net wrote: > >>This is an attempt to get the riscom8 driver working on SMP. I think I did it >>properly (per what Documentation/cli-sti-removal.txt details). > > > rc_interrupt also needs to protect data with locks (on UP it was ok, > because interrupts were disabled). > > Around line 1130 there are two msleep_interruptible calls under lock > (this was wrong before too). > > Some functions don't need to save flags, because you know they are > never called from interrupt (all from riscom_ops?). > > IMHO it would be nice to notice somewhere what is protected by lock > (rc_board and rc_port structures); and it would probably be better to > have a lock per board in riscom_board struct. > > > Domen > I had thought the same re putting the spinlock in the per-board structure (after submitting all of these - of course) - and got a lot of other pointers on how to make things SMP-correct. Go ahead and disregard this set of patches - I'll have to take a more careful look at each one of them. I thought I had seen some sleep-while-locked problems, but thought I was just reading it wrong. --===============7401475128657049== Content-Type: text/plain; charset="iso-8859-1" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline _______________________________________________ Kernel-janitors mailing list Kernel-janitors@lists.osdl.org http://lists.osdl.org/mailman/listinfo/kernel-janitors --===============7401475128657049==--