Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: [PATCH 11/18] pcmcia: do not use io_req_t when calling pcmcia_request_io()
Hi, http://userweb.kernel.org/~brodo/pcmcia-2.6.35.diff Serial part of Xircom card works with pcmcia-2.6.35.diff. - Original Message - Date: Sat, 18 Sep 2010 15:32:12 +0200 From: Dominik Brodowski li...@dominikbrodowski.net To: Komuro komurojun-...@nifty.com, pa...@ucw.cz Cc: linux-pcmcia@lists.infradead.org Subject: Re: Re: Re: Re: Re: Re: Re: Re: Re: [PATCH 11/18] pcmcia: do not use io_req_t when calling pcmcia_request_io() Hey, On Sat, Sep 18, 2010 at 09:50:31PM +0900, Komuro wrote: Also on re-inserting the card (with 2.6.36-rc4+ as of today)? I tried the 2.6.36-rc4-git4. Thanks for testing! elese we wouldn't see the is a 16550A message, right? Even if the COR(Configuration Option Register) of irq is not properly setup , we can see the above message... Nothing relating to setting CISREG_COR (at least on pcnet_cs) changed, and serial_cs does not modify CISREG_COR on such pseudo-multifunction devices. Would it be possible for you to re-check plain 2.6.35 and 2.6.35 with the pcmcia patchset availble at http://userweb.kernel.org/~brodo/pcmcia-2.6.35.diff ? If this works, we know that it's a serial/tty core issue, else it's indeed a PCMCIA issue. Pavel: has the bt / pcmcia issue you were seeing with -rc1 been solved? If not, could you try out 2.6.35 with this bigdiff as well, please? Thanks best, Dominik ___ Linux PCMCIA reimplementation list http://lists.infradead.org/mailman/listinfo/linux-pcmcia
Re: Re: Re: Re: Re: Re: Re: Re: Re: [PATCH 11/18] pcmcia: do not use io_req_t when calling pcmcia_request_io()
Hi! Nothing relating to setting CISREG_COR (at least on pcnet_cs) changed, and serial_cs does not modify CISREG_COR on such pseudo-multifunction devices. Would it be possible for you to re-check plain 2.6.35 and 2.6.35 with the pcmcia patchset availble at http://userweb.kernel.org/~brodo/pcmcia-2.6.35.diff ? If this works, we know that it's a serial/tty core issue, else it's indeed a PCMCIA issue. Pavel: has the bt / pcmcia issue you were seeing with -rc1 been solved? If not, could you try out 2.6.35 with this bigdiff as well, please? No, I still have the problem in -rc4. I did a quick test, and 2.6.35+your bigdiff works ok. (did not run make clean, that would be slow) Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html ___ Linux PCMCIA reimplementation list http://lists.infradead.org/mailman/listinfo/linux-pcmcia
Re: Re: Re: Re: Re: Re: Re: Re: [PATCH 11/18] pcmcia: do not use io_req_t when calling pcmcia_request_io()
Hi, With this patch, fmvj18x_cs based card works (both network and serial). but Serial part of Xircom card still does not work. I think the reason is irq or io_width or io_lines is not setuped properly for Xircom-card by pcmcia_core. Hey, On Sat, Sep 18, 2010 at 11:48:32AM +0900, Komuro wrote: Attached file is dmesg of another multi-function card(fmvj18x_cs) Thanks for the debug output -- this helps a lot! This card works with 2.6.35 but does not work with 2.6.36rc3(+patch). does this patch help (at least with the network function?)? Best, Dominik commit 9e7d4841d766f96a4e28a51c434485695d6a60ca Author: Dominik Brodowski li...@dominikbrodowski.net Date: Sat Sep 18 10:19:13 2010 +0200 pcmcia: preserve configuration information if request_io fails partly If pcmcia_request_io() only fails partly -- for the second of two requested resources -- preserve the configuration settings for the first one. Signed-off-by: Dominik Brodowski li...@dominikbrodowski.net diff --git a/drivers/pcmcia/pcmcia_resource.c b/drivers/pcmcia/pcmcia_resourc e.c index a5c1765..9ba4dad 100644 --- a/drivers/pcmcia/pcmcia_resource.c +++ b/drivers/pcmcia/pcmcia_resource.c @@ -595,7 +595,13 @@ int pcmcia_request_io(struct pcmcia_device *p_dev) if (c-io[1].end) { ret = alloc_io_space(s, c-io[1], p_dev-io_lines); if (ret) { + struct resource tmp = c-io[0]; + /* release the previously allocated resource */ release_io_space(s, c-io[0]); + /* but preserve the settings, for they worked... */ + c-io[0].end = resource_size(tmp); + c-io[0].start = tmp.start; + c-io[0].flags = tmp.flags; goto out; } } else ___ Linux PCMCIA reimplementation list http://lists.infradead.org/mailman/listinfo/linux-pcmcia
Re: Re: Re: Re: Re: Re: Re: Re: [PATCH 11/18] pcmcia: do not use io_req_t when calling pcmcia_request_io()
Hey, On Sat, Sep 18, 2010 at 07:20:31PM +0900, Komuro wrote: With this patch, fmvj18x_cs based card works (both network and serial). good. Will push the patch upstream, then. but Serial part of Xircom card still does not work. Also on re-inserting the card (with 2.6.36-rc4+ as of today)? I think the reason is irq or io_width or io_lines is not setuped properly for Xircom-card by pcmcia_core. But those values are the same for 2.6.35 and 2.6.36-rc4+, else we wouldn't see the same eth0: Xircom: port 0x300, irq 18, hwaddr 00:10:a4:f3:db:02 0.1: ttyS3 at I/O 0x2e8 (irq = 18) is a 16550A messages. Furthermore, at least some access to the serial part does work, elese we wouldn't see the is a 16550A message, right? Best, Dominik ___ Linux PCMCIA reimplementation list http://lists.infradead.org/mailman/listinfo/linux-pcmcia
Re: Re: Re: Re: Re: Re: Re: Re: Re: [PATCH 11/18] pcmcia: do not use io_req_t when calling pcmcia_request_io()
Hi, Also on re-inserting the card (with 2.6.36-rc4+ as of today)? I tried the 2.6.36-rc4-git4. The serial part still does not work, before and after re-inserting the card. elese we wouldn't see the is a 16550A message, right? Even if the COR(Configuration Option Register) of irq is not properly setup, we can see the above message... Hey, On Sat, Sep 18, 2010 at 07:20:31PM +0900, Komuro wrote: With this patch, fmvj18x_cs based card works (both network and serial). good. Will push the patch upstream, then. but Serial part of Xircom card still does not work. Also on re-inserting the card (with 2.6.36-rc4+ as of today)? I think the reason is irq or io_width or io_lines is not setuped properly for Xircom-card by pcmcia_core. But those values are the same for 2.6.35 and 2.6.36-rc4+, else we wouldn't see the same eth0: Xircom: port 0x300, irq 18, hwaddr 00:10:a4:f3:db:02 0.1: ttyS3 at I/O 0x2e8 (irq = 18) is a 16550A messages. Furthermore, at least some access to the serial part does work, elese we wouldn't see the is a 16550A message, right? Best, Dominik ___ Linux PCMCIA reimplementation list http://lists.infradead.org/mailman/listinfo/linux-pcmcia