From: Craig H. <cr...@gu...> - 2004-11-18 23:28:19
|
Ok folks, I think I might have nailed the BT problem. ------------------------------------------------------------------------ r234 | craig | 2004-11-18 15:14:07 -0800 (Thu, 18 Nov 2004) | 15 lines Committing a fix for the bluetooth problems in Mantis #35 The problem is that the serial link from the BT module to the BTUART was dropping bytes because the BT module does not respond quickly enough to RTS being dropped. There also appears to be a PXA bug wherein the BTUART is generating spurious interrupts without setting an interrupt source in the IIR, and th elinux driver was taking so long to respond to this condition that the non-spurious "read" interrupt was only happening and being processed after the buffer had overflowed. This problem is obviously exacerbated at higher baud rates. This fix drops RTS on entry into the ISR, processes all interrupts as though they might be read interrupts by reading the buffer until it's clear, and then re-asserting RTS only on exit from the ISR. This seems to work and allow reliable comms with the BT module at 921600 baud. ------------------------------------------------------------------------ Right now, this fix is on the modular-kernel branch. Could people who've seen bluetooth problems (which would be probably most people who have BT stix) grab r234 and build it, then test it on their stix? If I get a handful of "works for me"s then I'll merge the patch back onto the monolithic kernel branch and cut a new official root_fs release. This version also includes a few quick-and-dirty hacks to the bluetooth init script to reset the BT module before starting up, and to load all the required kernel modules before running the userspace apps. The fix is somewhat innelegant too, but is the only solution I could come up with to work around the various problems that exist (slow response to RTS by the ROK104001, spurious interrupts by the PXA, and slow ISR in linux). Note that this version *could* still fail, if there are a lot of other, non-BTUART interrupts happening in linux which prevent the BTUART ISR from kicking in in time to service the read buffer. As far as I can tell from the HCI H4 spec though, there is no way to recover from dropped bytes in the serial link, other than issuing an HCI reset command, and basically dumping all connections and all state, then re-establishing everything. I haven't implemented this yet, and the bluez H4 implementation seems to just completely ignore the issue of potential serial line errors (including not just overflows, but also parity and/or framing errors). I've also got a support ticket in to Infineon to see if they can think of any more graceful solution to this than what I've got now. Or if any of you can think of anything, I'd be happy to hear it too :) Let me know how r234 works for you, C |
From: Dave S. <da...@sm...> - 2004-11-21 13:15:37
|
On 18 Nov 2004, at 23:28, Craig Hughes wrote: > Ok folks, I think I might have nailed the BT problem. > > ----------------------------------------------------------------------- > - > r234 | craig | 2004-11-18 15:14:07 -0800 (Thu, 18 Nov 2004) | 15 lines > <snip> > Let me know how r234 works for you, > > C > OK, I'm lost. I've got the r234 root_fs_arm onto my mmc card, but I can't find the instructions on how to flash it. I know I've read them before, but I can't find it anymore on www.gumstix.org. Can someone please show me how inept I am and send me a link? :) Regards, Dave |
From: Craig H. <cr...@hu...> - 2004-11-21 16:17:27
|
On Nov 21, 2004, at 5:15 AM, Dave Smith wrote: > > On 18 Nov 2004, at 23:28, Craig Hughes wrote: > >> Ok folks, I think I might have nailed the BT problem. >> >> ---------------------------------------------------------------------- >> -- >> r234 | craig | 2004-11-18 15:14:07 -0800 (Thu, 18 Nov 2004) | 15 lines >> > <snip> >> Let me know how r234 works for you, >> >> C >> > > OK, I'm lost. I've got the r234 root_fs_arm onto my mmc card, but I > can't find the instructions on how to flash it. I know I've read them > before, but I can't find it anymore on www.gumstix.org. Can someone > please show me how inept I am and send me a link? :) Search the wiki for "fatload" C |
From: Dave S. <da...@sm...> - 2004-11-21 19:58:39
|
On 21 Nov 2004, at 16:11, Craig Hughes wrote: >> >> OK, I'm lost. I've got the r234 root_fs_arm onto my mmc card, but I >> can't find the instructions on how to flash it. I know I've read >> them before, but I can't find it anymore on www.gumstix.org. Can >> someone please show me how inept I am and send me a link? :) > > Search the wiki for "fatload" > > C > Thanks, found it, done it, and I think I've buggered up my waysmall :( I built it on my Mac using "make DEPMOD=true" as I don't have modprobe. However, on boot I now get could not parse modules.dep file And it does not exist in my filesystem. My /lib directory looks like # ls -laFR /lib /lib: drwxr-xr-x 3 root root 0 Nov 19 2004 ./ drwxr-xr-x 18 root root 0 Dec 31 16:00 ../ -rwxr-xr-x 1 root root 17328 Nov 19 2004 ld-uClibc-0.9.26.so* lrwxr-xr-x 1 root root 19 Nov 19 2004 ld-uClibc.so.0 -> ld-uClibc-0.9.26.so* lrwxr-xr-x 1 root root 19 Nov 19 2004 libc.so.0 -> libuClibc-0.9.26.so -rw-r--r-- 1 root root 10760 Nov 19 2004 libcrypt-0.9.26.so lrwxr-xr-x 1 root root 18 Nov 19 2004 libcrypt.so.0 -> libcrypt-0.9.26.so -rw-r--r-- 1 root root 6188 Nov 19 2004 libdl-0.9.26.so lrwxr-xr-x 1 root root 15 Nov 19 2004 libdl.so.0 -> libdl-0.9.26.so lrwxr-xr-x 1 root root 13 Nov 19 2004 libgcc_s.so -> libgcc_s.so.1 -rw-r--r-- 1 root root 38600 Nov 19 2004 libgcc_s.so.1 -rw-r--r-- 1 root root 50880 Nov 19 2004 libm-0.9.26.so lrwxr-xr-x 1 root root 14 Nov 19 2004 libm.so.0 -> libm-0.9.26.so -rw-r--r-- 1 root root 1500 Nov 19 2004 libnsl-0.9.26.so lrwxr-xr-x 1 root root 16 Nov 19 2004 libnsl.so.0 -> libnsl-0.9.26.so -rw-r--r-- 1 root root 68772 Nov 19 2004 libpthread-0.9.26.so lrwxr-xr-x 1 root root 20 Nov 19 2004 libpthread.so.0 -> libpthread-0.9.26.so -rw-r--r-- 1 root root 1500 Nov 19 2004 libresolv-0.9.26.so lrwxr-xr-x 1 root root 19 Nov 19 2004 libresolv.so.0 -> libresolv-0.9.26.so -rw-r--r-- 1 root root 263740 Nov 19 2004 libuClibc-0.9.26.so -rw-r--r-- 1 root root 4160 Nov 19 2004 libutil-0.9.26.so lrwxr-xr-x 1 root root 17 Nov 19 2004 libutil.so.0 -> libutil-0.9.26.so drwxr-xr-x 3 root root 0 Nov 19 2004 modules/ /lib/modules: drwxr-xr-x 3 root root 0 Nov 19 2004 ./ drwxr-xr-x 3 root root 0 Nov 19 2004 ../ drwxr-xr-x 3 root root 0 Nov 19 2004 2.6.9-rc1-gum/ /lib/modules/2.6.9-rc1-gum: drwxr-xr-x 3 root root 0 Nov 19 2004 ./ drwxr-xr-x 3 root root 0 Nov 19 2004 ../ drwxr-xr-x 2 root root 0 Nov 19 2004 kernel/ /lib/modules/2.6.9-rc1-gum/kernel: drwxr-xr-x 2 root root 0 Nov 19 2004 ./ drwxr-xr-x 3 root root 0 Nov 19 2004 ../ Which looks like a scary lack of modules. I've obviously got a broken build because I'm using a non-standard host. Can someone help me with a correctly built root_fs_arm for r234? Regards, Dave |
From: Craig H. <cr...@hu...> - 2004-11-21 20:16:07
|
Building one now for you. I'll post it somewhere "unofficial" to simplify things for those who've been having trouble building with the modular kernel. 10 minutes and it'll be done. C On Nov 21, 2004, at 11:56 AM, Dave Smith wrote: > > On 21 Nov 2004, at 16:11, Craig Hughes wrote: >>> >>> OK, I'm lost. I've got the r234 root_fs_arm onto my mmc card, but I >>> can't find the instructions on how to flash it. I know I've read >>> them before, but I can't find it anymore on www.gumstix.org. Can >>> someone please show me how inept I am and send me a link? :) >> >> Search the wiki for "fatload" >> >> C >> > > Thanks, found it, done it, and I think I've buggered up my waysmall :( > I built it on my Mac using "make DEPMOD=true" as I don't have > modprobe. However, on boot I now get > > could not parse modules.dep file > > And it does not exist in my filesystem. My /lib directory looks like > > # ls -laFR /lib > /lib: > drwxr-xr-x 3 root root 0 Nov 19 2004 ./ > drwxr-xr-x 18 root root 0 Dec 31 16:00 ../ > -rwxr-xr-x 1 root root 17328 Nov 19 2004 > ld-uClibc-0.9.26.so* > lrwxr-xr-x 1 root root 19 Nov 19 2004 ld-uClibc.so.0 > -> ld-uClibc-0.9.26.so* > lrwxr-xr-x 1 root root 19 Nov 19 2004 libc.so.0 -> > libuClibc-0.9.26.so > -rw-r--r-- 1 root root 10760 Nov 19 2004 > libcrypt-0.9.26.so > lrwxr-xr-x 1 root root 18 Nov 19 2004 libcrypt.so.0 > -> libcrypt-0.9.26.so > -rw-r--r-- 1 root root 6188 Nov 19 2004 libdl-0.9.26.so > lrwxr-xr-x 1 root root 15 Nov 19 2004 libdl.so.0 -> > libdl-0.9.26.so > lrwxr-xr-x 1 root root 13 Nov 19 2004 libgcc_s.so -> > libgcc_s.so.1 > -rw-r--r-- 1 root root 38600 Nov 19 2004 libgcc_s.so.1 > -rw-r--r-- 1 root root 50880 Nov 19 2004 libm-0.9.26.so > lrwxr-xr-x 1 root root 14 Nov 19 2004 libm.so.0 -> > libm-0.9.26.so > -rw-r--r-- 1 root root 1500 Nov 19 2004 > libnsl-0.9.26.so > lrwxr-xr-x 1 root root 16 Nov 19 2004 libnsl.so.0 -> > libnsl-0.9.26.so > -rw-r--r-- 1 root root 68772 Nov 19 2004 > libpthread-0.9.26.so > lrwxr-xr-x 1 root root 20 Nov 19 2004 > libpthread.so.0 -> libpthread-0.9.26.so > -rw-r--r-- 1 root root 1500 Nov 19 2004 > libresolv-0.9.26.so > lrwxr-xr-x 1 root root 19 Nov 19 2004 libresolv.so.0 > -> libresolv-0.9.26.so > -rw-r--r-- 1 root root 263740 Nov 19 2004 > libuClibc-0.9.26.so > -rw-r--r-- 1 root root 4160 Nov 19 2004 > libutil-0.9.26.so > lrwxr-xr-x 1 root root 17 Nov 19 2004 libutil.so.0 > -> libutil-0.9.26.so > drwxr-xr-x 3 root root 0 Nov 19 2004 modules/ > > /lib/modules: > drwxr-xr-x 3 root root 0 Nov 19 2004 ./ > drwxr-xr-x 3 root root 0 Nov 19 2004 ../ > drwxr-xr-x 3 root root 0 Nov 19 2004 2.6.9-rc1-gum/ > > /lib/modules/2.6.9-rc1-gum: > drwxr-xr-x 3 root root 0 Nov 19 2004 ./ > drwxr-xr-x 3 root root 0 Nov 19 2004 ../ > drwxr-xr-x 2 root root 0 Nov 19 2004 kernel/ > > /lib/modules/2.6.9-rc1-gum/kernel: > drwxr-xr-x 2 root root 0 Nov 19 2004 ./ > drwxr-xr-x 3 root root 0 Nov 19 2004 ../ > > Which looks like a scary lack of modules. I've obviously got a broken > build because I'm using a non-standard host. Can someone help me with > a correctly built root_fs_arm for r234? > > Regards, > Dave > > > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real > users. > Discover which products truly live up to the hype. Start reading now. > http://productguide.itmanagersjournal.com/ > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: Craig H. <cr...@hu...> - 2004-11-21 20:41:24
|
Ok, I've put a root_fs_arm from r234 on http://azazel.rungie.com/~craig/bt-beta/root_fs_arm C |
From: Dave S. <da...@sm...> - 2004-11-21 20:56:06
|
On 21 Nov 2004, at 20:39, Craig Hughes wrote: > Ok, I've put a root_fs_arm from r234 on > http://azazel.rungie.com/~craig/bt-beta/root_fs_arm > > C > > Thanks so much, I'm downloading it now. Regards, Dave |
From: Lawrence H. <lh...@ha...> - 2004-12-01 10:06:19
|
I may just be blind but I have looked through the wiki and sourceforge and I can't find r234 to load it. Can someone tell me where it is. When I try the bluetooth on my gumstix I see it go up and down like a yoyo which I think is the symptom fixed by the patches in r234. Thanks, Lawrence Harris ----- Original Message ----- From: "Craig Hughes" <cr...@hu...> To: <gum...@li...> Sent: Sunday, November 21, 2004 8:11 AM Subject: Re: [Gumstix-users] BT problems fix > On Nov 21, 2004, at 5:15 AM, Dave Smith wrote: > >> >> On 18 Nov 2004, at 23:28, Craig Hughes wrote: >> >>> Ok folks, I think I might have nailed the BT problem. >>> >>> ---------------------------------------------------------------------- >>> -- >>> r234 | craig | 2004-11-18 15:14:07 -0800 (Thu, 18 Nov 2004) | 15 lines >>> >> <snip> >>> Let me know how r234 works for you, >>> >>> C >>> >> >> OK, I'm lost. I've got the r234 root_fs_arm onto my mmc card, but I >> can't find the instructions on how to flash it. I know I've read them >> before, but I can't find it anymore on www.gumstix.org. Can someone >> please show me how inept I am and send me a link? :) > > Search the wiki for "fatload" > > C > > > > ------------------------------------------------------- > This SF.Net email is sponsored by: InterSystems CACHE > FREE OODBMS DOWNLOAD - A multidimensional database that combines > robust object and relational technologies, making it a perfect match > for Java, C++,COM, XML, ODBC and JDBC. www.intersystems.com/match8 > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > |
From: Craig H. <cr...@hu...> - 2004-12-01 18:17:51
|
Lawrence, r234 refers to a subversion revision number. There are 2 ways of getting the code. First is to use subversion to check that revision out, then compile it. The other way is to fetch the correct revision from http://svn.rungie.com/websvn/ C On Dec 1, 2004, at 2:06 AM, Lawrence Harris wrote: > I may just be blind but I have looked through the wiki and sourceforge > and I can't find r234 to load it. Can someone tell me where it is. > When I try the bluetooth on my gumstix I see it go up and down like a > yoyo which I think is the symptom fixed by the patches in r234. > > Thanks, Lawrence Harris > > ----- Original Message ----- From: "Craig Hughes" > <cr...@hu...> > To: <gum...@li...> > Sent: Sunday, November 21, 2004 8:11 AM > Subject: Re: [Gumstix-users] BT problems fix > > >> On Nov 21, 2004, at 5:15 AM, Dave Smith wrote: >> >>> >>> On 18 Nov 2004, at 23:28, Craig Hughes wrote: >>> >>>> Ok folks, I think I might have nailed the BT problem. >>>> >>>> -------------------------------------------------------------------- >>>> -- -- >>>> r234 | craig | 2004-11-18 15:14:07 -0800 (Thu, 18 Nov 2004) | 15 >>>> lines >>>> >>> <snip> >>>> Let me know how r234 works for you, >>>> >>>> C >>>> >>> >>> OK, I'm lost. I've got the r234 root_fs_arm onto my mmc card, but I >>> can't find the instructions on how to flash it. I know I've read >>> them before, but I can't find it anymore on www.gumstix.org. Can >>> someone please show me how inept I am and send me a link? :) >> >> Search the wiki for "fatload" >> >> C >> >> >> >> ------------------------------------------------------- >> This SF.Net email is sponsored by: InterSystems CACHE >> FREE OODBMS DOWNLOAD - A multidimensional database that combines >> robust object and relational technologies, making it a perfect match >> for Java, C++,COM, XML, ODBC and JDBC. www.intersystems.com/match8 >> _______________________________________________ >> gumstix-users mailing list >> gum...@li... >> https://lists.sourceforge.net/lists/listinfo/gumstix-users > > > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real > users. > Discover which products truly live up to the hype. Start reading now. > http://productguide.itmanagersjournal.com/ > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: Dave S. <da...@sm...> - 2004-11-21 22:24:30
|
On 18 Nov 2004, at 23:28, Craig Hughes wrote: > > > Let me know how r234 works for you, > > C > mmc support appears to be borked. # modprobe pxamci # Allocated rca: 0001 # modprobe mmcblock modprobe: module mmcblock not found. modprobe: failed to load module mmcblock # modprobe vfat # modprobe nls_cp437 # modprobe nls_iso8859-1 # mount /mnt/mmc mount: Mounting /dev/mmc/mmc0/part1 on /mnt/mmc failed: No such device or address # pwd /dev/mmc/mmc0 # ls disc part1 # ls -laF drwxr-xr-x 2 root root 0 Nov 21 2004 ./ drwxr-xr-x 3 root root 0 Nov 21 2004 ../ brw-r----- 1 root root 254, 0 Nov 21 2004 disc brw-r----- 1 root root 254, 1 Nov 21 2004 part1 I guess this is just a symptom of the modular kernel being in flux right now. |
From: Dave S. <da...@sm...> - 2004-11-21 22:07:46
|
On 18 Nov 2004, at 23:28, Craig Hughes wrote: > > Right now, this fix is on the modular-kernel branch. Could people > who've seen bluetooth problems (which would be probably most people > who have BT stix) grab r234 and build it, then test it on their stix? > If I get a handful of "works for me"s then I'll merge the patch back > onto the monolithic kernel branch and cut a new official root_fs > release. > > Thanks for the build :) OK, on first boot from the u-boot prompt, bluetooth still seemed ropy, I was getting invalid tx (? something like that, it's long since scrolled off my screen :( ) errors on any bluetooth usage. In best tech support mode, I re-booted and it now seems fine. I've paired it with my Mac (which I was never able to do before), and it hasn't dropped out at all. It paired fine with my Palm Tungsten. Probably a config thing, but I haven't been able to browse its services yet, from my BSD box if I run sdpcontrol -a gs_hex_addr browse I get "Could not execute command 'browse' Socket not connected", but if I do the same pointing at the Mac I get a full list of capabilities. I probably need to dive into /etc/bluetooth, google and Bluez docs. My /var/log/messages file is getting stuffed with a udhcpc sending discover message approx every 4 seconds. Regards, Dave |
From: Florian L. <gu...@fl...> - 2004-11-21 23:24:50
|
On Sunday 21 November 2004 23:06, Dave Smith wrote: > On 18 Nov 2004, at 23:28, Craig Hughes wrote: > > Right now, this fix is on the modular-kernel branch. Could people > > who've seen bluetooth problems (which would be probably most people > > who have BT stix) grab r234 and build it, then test it on their stix? > > If I get a handful of "works for me"s then I'll merge the patch back > > onto the monolithic kernel branch and cut a new official root_fs > > release. > no luck here :( connection is establishing (using pand), but I can't use the connection. Even a simple l2ping would result in the following error-messages: hci_acl_tx_to: hci0 ACL tx timeout hci_acl_tx_to: hci0 killing stalled ACL connection CE:79:00:94:0A:00 ... [repeated] I'm having a GS 200Mhz. // florian -- This function will terminate, if run infinitely. void f() { while (random() != 0); } |