From: song g. <guo...@gm...> - 2007-02-15 20:49:46
|
How could I know my netstix is 400 or 200 thank you song |
From: Craig H. <cr...@gu...> - 2007-02-15 20:57:36
|
On Feb 15, 2007, at 12:49 PM, song guo wrote: > How could I know my netstix is 400 or 200 If you haven't reflashed the u-boot on it, then you can check /proc/ cpuinfo in linux. The "bogomips" number is approximately the megaherz rate on the PXA255/270. If you *have* reflashed u-boot and might have flashed the wrong speed u-boot onto the board, then the only way to tell is to open the case and read the label on the CPU package. It will say something like "PXA255A0C200" or something. Those last 3 digits are the speed rating of the CPU. C |
From: DJ D. <dj...@de...> - 2007-02-15 21:04:09
|
Craig Hughes <cr...@gu...> writes: > package. It will say something like "PXA255A0C200" or something. > Those last 3 digits are the speed rating of the CPU. On a related note, is it possible to make a 400 run at 200MHz for testing? I got a 400 for prototyping, but for the dedicated one I might as well buy a 200 if it's fast enough, but I'd like to "try and see" first. |
From: Julien L. <les...@ho...> - 2007-02-15 23:45:56
|
Yep it is possible, just include the frequence scaling option in the gumstix linux kernel, then it'll be available under /sys/devices/system/cpu/cpu0/cpufreq/ Here you'll find a bunch of files, the details for each can be find in the kernel doc in the directory /cpufreq, but to downclock simply do: echo 200000 > scaling_min_freq // set the min freq for the clock echo 200000 > scaling_set_speed // set the clock to 200Mhz check by doing cat scaling_set_speed it should give 200000 the cpuinfo_min_freq and cpuinfo_max_freq give the min/max value of the clock. Note that you can also overclock the cpu this way, I've run my gumstix to 530Mhz but for the moment it's useless ;) Julien >On a related note, is it possible to make a 400 run at 200MHz for >testing? I got a 400 for prototyping, but for the dedicated one I >might as well buy a 200 if it's fast enough, but I'd like to "try and >see" first. > _________________________________________________________________ Testez Windows Llive Mail Beta ! http://www.msn.fr/newhotmail/Default.asp?Ath=f |
From: DJ D. <dj...@de...> - 2007-02-15 23:58:52
|
Cool. I don't have that at the moment, but I'll keep it in mind for when I get to that point. Thanks! |
From: Julien L. <les...@ho...> - 2007-02-16 00:00:36
|
Hi, Thanks for all the help, it does compile fine now ! In fact everything was right from the beginning but it was just the UC1400 touch-screen which was crashing during the compilation. I couldn't figure out at first because make was printing the error for the touchscreen and crashed while compiling the things under ipv4 ... So I don't know why but just by removing the UC1400 touchscreen support it works... But how can I use the touchscreen then :) ? Another thing, I've imported my programs I've wrote with the previous buildroot and the user-mode programs just compile finely but my kernel driver gives strange errors: lesnip3r@A6J-Julien:~/gumstix-buildroot/build_arm_nofpu/uLCD_drv$ make make -C /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/linux-2.6.20gum M=/home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv ARCH=arm CROSS_COMPILE=/home/lesnip3r/gumstix-buildroot/build_arm_nofpu/staging_dir/bin/arm-linux- \ CC=/home/lesnip3r/gumstix-buildroot/build_arm_nofpu/staging_dir/bin/arm-linux-gcc modules make[1]: entrant dans le répertoire « /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/linux-2.6.20gum » CC [M] /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.o In file included from include/linux/jiffies.h:4, from include/linux/sched.h:51, from include/linux/smp_lock.h:5, from include/linux/hardirq.h:5, from include/asm-generic/local.h:5, from include/asm/local.h:1, from include/linux/module.h:19, from /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.c:3: include/linux/calc64.h: In function 'do_div_llr': include/linux/calc64.h:25: error: '__LINUX_ARM_ARCH__' undeclared (first use in this function) include/linux/calc64.h:25: error: (Each undeclared identifier is reported only once include/linux/calc64.h:25: error: for each function it appears in.) In file included from include/linux/aio.h:5, from include/linux/sched.h:417, from include/linux/smp_lock.h:5, from include/linux/hardirq.h:5, from include/asm-generic/local.h:5, from include/asm/local.h:1, from include/linux/module.h:19, from /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.c:3: include/linux/workqueue.h: In function 'cancel_delayed_work': include/linux/workqueue.h:206: warning: dereferencing type-punned pointer will break strict-aliasing rules include/linux/workqueue.h:206: warning: dereferencing type-punned pointer will break strict-aliasing rules /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.c:5:27: error: linux/config.h: No such file or directory In file included from /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.c:15: /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/shakira.h:16144:3: warning: no newline at end of file make[2]: *** [/home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.o] Erreur 1 make[1]: *** [_module_/home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv] Erreur 2 make[1]: quittant le répertoire « /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/linux-2.6.20gum » make: *** [uLCD.ko] Erreur 2 For the linux/config.h I've read the answer earlier today so it should be fine but for the rest... May it be because I have not adrressed the linux/config.h issue ? I believe my makefile to be good, and the errors do not show my program as faulty... What happened ? Julien _________________________________________________________________ MSN Messenger : discutez en direct avec vos amis ! http://www.msn.fr/msger/default.asp |
From: Craig H. <cr...@gu...> - 2007-02-16 00:32:47
|
On Feb 15, 2007, at 3:45 PM, Julien Lebot wrote: > Yep it is possible, > just include the frequence scaling option in the gumstix linux > kernel, then > it'll be available under /sys/devices/system/cpu/cpu0/cpufreq/ > > Here you'll find a bunch of files, the details for each can be find > in the > kernel doc in the directory /cpufreq, but to downclock simply do: > echo 200000 > scaling_min_freq // set the min freq for the clock > echo 200000 > scaling_set_speed // set the clock to 200Mhz > > check by doing > cat scaling_set_speed > it should give 200000 > > the cpuinfo_min_freq and cpuinfo_max_freq give the min/max value of > the > clock. > Note that you can also overclock the cpu this way, I've run my > gumstix to > 530Mhz but for the moment it's useless ;) Note that there's a pretty serious PXA255 erratum which means that sometimes when you switch CPU frequencies, just about all the device controllers on the PXA lose state and need to be reset; the linux drivers do not do this. I've only ever seen this show up when switching frequencies *a lot* using the "ondemand" cpufreq driver. Single manual switches should be fine (though not guaranteed to be so afaik). The safest way to test a gumstix at a different speed is to use the other-speed u-boot, which sets the CPU core frequency before initializing any of the device controllers. The erratum is gone on the PXA270, which switches frequencies using the ondemand driver very nicely (clocking down to 13MHz when very idle, saves lots of juice!) C |
From: Dave H. <dhy...@gm...> - 2007-02-16 00:43:09
|
Hi Julien, > Thanks for all the help, it does compile fine now ! In fact everything was > right from the beginning but it was just the UC1400 touch-screen which was > crashing during the compilation. I couldn't figure out at first because make > was printing the error for the touchscreen and crashed while compiling the > things under ipv4 ... This is because the default make is run with -j 3 which causes 3 compiles to be running in parallel. > So I don't know why but just by removing the UC1400 touchscreen support it > works... But how can I use the touchscreen then :) ? You'd need to add the proper support to make it work. It's probably the case that this hasn't been done yet. > Another thing, I've imported my programs I've wrote with the previous > buildroot and the user-mode programs just compile finely but my kernel > driver gives strange errors: What does your Makefile look like for the kernel modules? > include/linux/calc64.h:25: error: '__LINUX_ARM_ARCH__' undeclared (first use > in this function) For my drivers, the kernel Makefile automatically adds a -D option for __LINUX_ARM_ARCH__ > /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.c:5:27: > error: linux/config.h: No such file or directory include/linux/config.h no longer exists (it was removed in 2.6.19). You can try to use <linux/autoconf.h> instead. > In file included from > /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/uLCD.c:15: > /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv/shakira.h:16144:3: > warning: no newline at end of file This is just an ANSI C warning. All lines must end in a newline, and the last line of shakira.h has some text but no newline at the end. Add a newline at the end of the file to get rid of the warning. > For the linux/config.h I've read the answer earlier today so it should be > fine but for the rest... May it be because I have not adrressed the > linux/config.h issue ? > I believe my makefile to be good, and the errors do not show my program as > faulty... > What happened ? I'd be happy to review your Makefile. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Craig H. <cr...@gu...> - 2007-02-16 01:12:14
|
On Feb 15, 2007, at 4:31 PM, Dave Hylands wrote: >> So I don't know why but just by removing the UC1400 touchscreen >> support it >> works... But how can I use the touchscreen then :) ? > > You'd need to add the proper support to make it work. It's probably > the case that this hasn't been done yet. Hmm, it should be working -- I had it working in 2.6.18; could be that 2.6.20 broke something? C |
From: Julien L. <les...@ho...> - 2007-02-16 10:10:36
|
Hi, >This is because the default make is run with -j 3 which causes 3 >compiles to be running in parallel. Ok I see, no big stuff, I just have to be careful looking in the history for the real errors :) >You'd need to add the proper support to make it work. It's probably >the case that this hasn't been done yet. You mean writing my own driver ? It looks like it has been designed to work with the gumstix because with the other platforms the UC1400 doesn't show up... May I should try debugging it, but I'm stil noob, nobody else use the UC1400 touchscreen there ?? > >What does your Makefile look like for the kernel modules? ############################################################################# # # Makefile for building the char-driver sample module # ############################################################################# ifeq ($(DEBUG),y) CFLAGS += -O -g # -O is need to expand inlines else CFLAGS += -O2 endif TARGET = uLCD TARGET_ARCH = arm LINUX_VERSION = 2.6.20gum PWD := $(shell pwd) GUMSTIX_BUILDROOT = $(PWD)/../gumstix-buildroot BUILD_ARM = $(wildcard $(GUMSTIX_BUILDROOT)/build_arm*) KERNELDIR = $(wildcard $(BUILD_ARM)/linux-$(LINUX_VERSION) ) CROSS_COMPILE = /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/staging_dir/bin/arm-linux- LINUX_DIR = /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/linux-$(LINUX_VERSION) srcdir = /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/uLCD_drv CC = $(CROSS_COMPILE)gcc CPP = /home/lesnip3r/gumstix-buildroot/build_arm_nofpu/staging_dir/bin/arm-linux-uclibcgnueabi-cpp0 CFLAGS = -Wall -O2 MODULES = uLCD HOST_BINARIES = TARGET_BINARIES = default: all # More rules for the development version -include $(SUBDIRS)/Makefile.priv # When called from kbuild -include Makefile.priv # Other calls obj-m += $(MODULES:%=%.o) BUILD = $(MODULES:%=%.ko) \ $(HOST_BINARIES:%=%.host) \ $(TARGET_BINARIES:%=%.target) all: $(BUILD) clean: rm -f $(BUILD) *.o *.ko *.mod.c *.mod.o *~ .*.cmd Module.symvers rm -rf .tmp_versions $(MODULES:%=%.ko): *.c *.h $(MAKE) -C $(LINUX_DIR) M=$(PWD) ARCH=$(TARGET_ARCH) CROSS_COMPILE=$(CROSS_COMPILE) \ CC=$(CC) modules %.target: %.c *.h $(CC) $(CFLAGS) -o $@ $< ########################################################## In fact it's my makefile for everything, all I have to do is specify the targets. It's a compilation of the makefiles I found on the wiki and in the sources. > >For my drivers, the kernel Makefile automatically adds a -D option for >__LINUX_ARM_ARCH__ > I have to add this option to make ? > >include/linux/config.h no longer exists (it was removed in 2.6.19). >You can try to use <linux/autoconf.h> instead. > Yep I've seen your mail this morning, I'll do that. > >This is just an ANSI C warning. All lines must end in a newline, and >the last line of shakira.h has some text but no newline at the end. >Add a newline at the end of the file to get rid of the warning. > Yeah of course :D I've always ignored this error, it's just a computer generated picture of shakira stored in a .h file so no big stuff here too :) Again thanks for the help, I'm making huge progress here ;) Julien _________________________________________________________________ Testez Windows Llive Mail Beta ! http://www.msn.fr/newhotmail/Default.asp?Ath=f |
From: Dave H. <dhy...@gm...> - 2007-02-16 14:52:50
|
Hi Julien, > You mean writing my own driver ? It looks like it has been designed to work > with the gumstix because with the other platforms the UC1400 doesn't show > up... May I should try debugging it, but I'm stil noob, nobody else use the > UC1400 touchscreen there ?? I haven't used the touchscreen driver before, so I was just speculating. Craig says it worked until just recently, so it probably just needs some updates or something. > >What does your Makefile look like for the kernel modules? OK - here's what I do for makefiles which I want to make kernel stuff with: > ############################################################################# > # > # Makefile for building the char-driver sample module > # > ############################################################################# > > ifeq ($(DEBUG),y) > CFLAGS += -O -g # -O is need to expand inlines > else > CFLAGS += -O2 > endif Everything up tp here is fine. Then I put this in: ifneq ($(KERNELRELEASE),) # call from kernel build system obj-m := single-object file for module else Now the rest of the Makefle can go in the else portion with an endif at the end of the file. What's really significant is that the kernel Makefile "includes" your Makefile, so anything that you do (assigning variables, creating rules) on the kernel side of things can mess stuff up. This isn't quite so bad when you're makeing a single module outside the kernel tree, but can have significant ramifications when building inside the kernel tree. > $(MODULES:%=%.ko): *.c *.h > $(MAKE) -C $(LINUX_DIR) M=$(PWD) ARCH=$(TARGET_ARCH) > CROSS_COMPILE=$(CROSS_COMPILE) \ > CC=$(CC) modules This should be changed to look like this: $(MODULES:%=%.ko): *.c *.h $(MAKE) -C $(LINUX_DIR) M=$(PWD) ARCH=$(TARGET_ARCH) CROSS_COMPILE=$(CROSS_COMPILE) modules It's the CC= portion that's messing things up. It's causing the __LINUX_ARM_ARCH__ define in the linux makefile to get removed. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Julien L. <les...@ho...> - 2007-02-17 10:11:04
|
Hi ! Thanks, everything is working finely now, well at least it compiles :) Now I just have to compile my modules like kernel modules, but is there anyway for me to skip typing: make ARCH=arm DEPMOD=/bin/true CROSS_COMPILE="`pwd`/../staging_dir/bin/arm-linux-" each time I want to build my module ? (If I don't do that, it won't compile... and all these informations are included in my makefile.... weird) Anyway thanks, I'll do a big backup and start programming again now that I'm not limited by the compilation process. Julien _________________________________________________________________ MSN Messenger: appels gratuits de PC à PC ! http://www.msn.fr/newhotmail/Default.asp?Ath=f |
From: Dave H. <dhy...@gm...> - 2007-02-17 17:43:03
|
Hi Julien, On 2/17/07, Julien Lebot <les...@ho...> wrote: > Hi ! > > Thanks, everything is working finely now, well at least it compiles :) > Now I just have to compile my modules like kernel modules, but is there > anyway for me to skip typing: > > make ARCH=arm DEPMOD=/bin/true > CROSS_COMPILE="`pwd`/../staging_dir/bin/arm-linux-" No need to type all that just to build your module. Your makefile already does it for you. This line in your Makefile: $(MODULES:%=%.ko): *.c *.h $(MAKE) -C $(LINUX_DIR) M=$(PWD) ARCH=$(TARGET_ARCH) CROSS_COMPILE=$(CROSS_COMPILE) modules The only thing it seems to be missing is the DEPMOD=/bin/true If you add that, then you can just type make and your makefile will in turn call the kernel make with all the right options. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Jesse W. <jes...@gm...> - 2007-02-17 10:48:45
|
if you are using bash you can just say something like: MKSTRING=3D"what ever you want" make $MKSTRING but you will have to look carefully into all those """''' and make sure they are themselves controlled so they don't end your string. this should help http://tldp.org/HOWTO/Bash-Prog-Intro-HOWTO.html On 2/17/07, Julien Lebot <les...@ho...> wrote: > Hi ! > > Thanks, everything is working finely now, well at least it compiles :) > Now I just have to compile my modules like kernel modules, but is there > anyway for me to skip typing: > > make ARCH=3Darm DEPMOD=3D/bin/true > CROSS_COMPILE=3D"`pwd`/../staging_dir/bin/arm-linux-" > > each time I want to build my module ? (If I don't do that, it won't > compile... and all these informations are included in my makefile.... weird) > > Anyway thanks, I'll do a big backup and start programming again now that I'm > not limited by the compilation process. > > Julien > > _________________________________________________________________ > MSN Messenger: appels gratuits de PC =E0 PC ! > http://www.msn.fr/newhotmail/Default.asp?Ath=3Df > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=3Djoin.php&p=3Dsourceforge&CID=3D= DEVDEV > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > --=20 -Jesse W. |
From: Craig H. <cr...@gu...> - 2007-02-18 21:59:38
|
I do: alias amake=3D'make ARCH=3Darm CROSS_COMPILE=3D"`pwd`/../staging_dir/bin/=20= arm-linux-"' I've never needed the DEPMOD=3D thing. C On Feb 17, 2007, at 2:48 AM, Jesse Welling wrote: > if you are using bash you can just say something like: > > MKSTRING=3D"what ever you want" > make $MKSTRING > > but you will have to look carefully into all those """''' and make =20= > sure they are themselves controlled so they don't end your string. =20 > this should help http://tldp.org/HOWTO/Bash-Prog-Intro-HOWTO.html > > On 2/17/07, Julien Lebot <les...@ho...> wrote: > > Hi ! > > > > Thanks, everything is working finely now, well at least it =20 > compiles :) > > Now I just have to compile my modules like kernel modules, but is =20= > there > > anyway for me to skip typing: > > > > make ARCH=3Darm DEPMOD=3D/bin/true > > CROSS_COMPILE=3D"`pwd`/../staging_dir/bin/arm-linux-" > > > > each time I want to build my module ? (If I don't do that, it won't > > compile... and all these informations are included in my =20 > makefile.... weird) > > > > Anyway thanks, I'll do a big backup and start programming again =20 > now that I'm > > not limited by the compilation process. > > > > Julien > > > > _________________________________________________________________ > > MSN Messenger: appels gratuits de PC =E0 PC ! > > http://www.msn.fr/newhotmail/Default.asp?Ath=3Df > > > > > > =20 > ----------------------------------------------------------------------=20= > --- > > Take Surveys. Earn Cash. Influence the Future of IT > > Join SourceForge.net's Techsay panel and you'll get the chance to =20= > share your > > opinions on IT & business topics through brief surveys-and earn cash > > http://www.techsay.com/default.php?=20 > page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV > > _______________________________________________ > > gumstix-users mailing list > > gum...@li... > > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > > > > --=20 > -Jesse W. > ----------------------------------------------------------------------=20= > --- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to =20 > share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?=20 > page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV____________________________= ____=20 > _______________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: Julien L. <les...@ho...> - 2007-02-23 06:23:14
|
Hi all, Some news from the front :) I've been able to compile the touch-screen module. Here is what happened, when compiling I've just copy/pasted this: make ARCH=arm DEPMOD=/bin/true CROSS_COMPILE="`pwd`/../staging_dir/ bin/arm-linux-" menuconfig from the mail Craig wrote, but linux put a return carriage after /staging_dir/ blowing any chance for me to build the kernel... When I've been able to compile (and not being stopped at the first line of the first module), a couple of errors showed up concerning the UCB1400 touch-screen (as usual...). I've looked in UC1xxx-core.c for the errors, and naively deleted the faulty lines: - a reference to try_to_freeze -a line with a mismatched pointer Everything is on the linux side of my laptop, so if anybody wants more info tell me because that would imply reproducing the errors :p Now it's compiled, and I've access to the functions of the UCB1400 (although I haven't tested yet, they show up on my gumstix). > > You mean writing my own driver ? It looks like it has been designed to >work > > with the gumstix because with the other platforms the UC1400 doesn't >show > > up... May I should try debugging it, but I'm stil noob, nobody else use >the > > UC1400 touchscreen there ?? Ok now a couple of nasty question (sorry, noob inside): First, can I transfer only the kernel to the gumstix without having to flash it ? Because I don't touch anymore the package selection, just the kernel configuration. Secondly: "You have part of the driver compiled into the kernel, and part as a module. I would recommend making everything be compiled in. So all those things that are "=m" need to be "=y". Let's see if that helps. Steve" I've compiled with everything as part of the kernel (not module), the audio, the framebuffer, even the GPIO... Now what happen is that I still don't have the framebuffer up and running, the module doesn't exist anymore (that's normal I guess), I got 2 gpio folders in my /dev one is for the UCB1400 and the other I suppose it's for the gumstix, but I can't access the gumstix's one now.... >From the bootlog, I see nothing related to the framebuffer or pxa2xx-fb... And also it says something about the gpio module not found (guess it's normal too, I'll just have to remove the module name from /etc/modules). So any idea ? Thanks in advance, Julien _________________________________________________________________ MSN Hotmail : créez votre adresse e-mail gratuite & à vie ! http://www.msn.fr/newhotmail/Default.asp?Ath=f |
From: Dave H. <dhy...@gm...> - 2007-02-23 06:28:18
|
Hi Steve, > Ok now a couple of nasty question (sorry, noob inside): > > First, can I transfer only the kernel to the gumstix without having to flash > it ? Because I don't touch anymore the package selection, just the kernel > configuration. Provided that the kernel that was there came from your buildroot, and you have enough space (preferably an XM), then yeah you can do it. You could remove the old uImage and copy the new one in. Worst case, is you need to download the whole thing. If you compiled the driver as a module, then you just need to copy the module over, and you can leave the kernel alone. For developing a new driver, I really like having it as a module, cause you can load and unload and if it really screws up you just reboot. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Julien L. <les...@ho...> - 2007-02-23 06:47:55
|
Hi Dave, That's me, Julien, who posted the question, just to make sure nobody gets confused, I'm not talking for Steve :) >Hi Steve, > > > Ok now a couple of nasty question (sorry, noob inside): > > > > First, can I transfer only the kernel to the gumstix without having to >flash > > it ? Because I don't touch anymore the package selection, just the >kernel > > configuration. > >Provided that the kernel that was there came from your buildroot, and >you have enough space (preferably an XM), then yeah you can do it. > >You could remove the old uImage and copy the new one in. Worst case, >is you need to download the whole thing. > >If you compiled the driver as a module, then you just need to copy the >module over, and you can leave the kernel alone. > >For developing a new driver, I really like having it as a module, >cause you can load and unload and if it really screws up you just >reboot. > >-- I have to agree with you, when developing it's useful, if it doesn't work, unload and upload a new version. But it's about the framebuffer, it just doesn't show up... But I recall now that after compiling the framebuffer as a part of the kernel, I have a file "fb" in my /proc folder... may it be the Graal I'm looking for ?? Then how do I access it from the kernel driver ? Thanks for your answer, Julien _________________________________________________________________ MSN Messenger : discutez en direct avec vos amis ! http://www.msn.fr/msger/default.asp |
From: Alexis C. <chi...@ya...> - 2007-02-23 19:27:20
Attachments:
log.txt
|
Julien, Any progress on the framebuffer side ?? I think I have the same issues... Framebuffer driver is built in the kernel : # # Graphics support # # CONFIG_FIRMWARE_EDID is not set CONFIG_FB=y CONFIG_FB_CFB_FILLRECT=y CONFIG_FB_CFB_COPYAREA=y CONFIG_FB_CFB_IMAGEBLIT=y # CONFIG_FB_MACMODES is not set # CONFIG_FB_BACKLIGHT is not set CONFIG_FB_MODE_HELPERS=y # CONFIG_FB_TILEBLITTING is not set # CONFIG_FB_S1D13XXX is not set CONFIG_FB_PXA=y CONFIG_FB_PXA_PARAMETERS=y # CONFIG_FB_MBX is not set # CONFIG_FB_VIRTUAL is not set While it is booting I get this : pxa2xx-fb: probe of pxa2xx-fb failed with error -12 There is nothing before or after in the log that can help, but you can find the full bootlog attached. I am using distrib release 1295 and my config is a basix 400 XM with breakout GS. BTW, nothing yet is connected to the LCD pins on the breakout-GS, but I don't think this is the issue. Of course, I cannot see the /dev/fb0, I only get the /proc/fb. I tried as a module, and it gives me the same error. Any help would be appreciated, and I don't seem to be the only one in that situation. Best, Alexis. ----- Original Message ----- From: "Julien Lebot" <les...@ho...> To: <gum...@li...> Sent: Friday, February 23, 2007 1:47 AM Subject: Re: [Gumstix-users] About the gumstix kernel / UCB1400 / Framebuffer > Hi Dave, > > That's me, Julien, who posted the question, just to make sure nobody gets > confused, I'm not talking for Steve :) > >>Hi Steve, >> >> > Ok now a couple of nasty question (sorry, noob inside): >> > >> > First, can I transfer only the kernel to the gumstix without having to >>flash >> > it ? Because I don't touch anymore the package selection, just the >>kernel >> > configuration. >> >>Provided that the kernel that was there came from your buildroot, and >>you have enough space (preferably an XM), then yeah you can do it. >> >>You could remove the old uImage and copy the new one in. Worst case, >>is you need to download the whole thing. >> >>If you compiled the driver as a module, then you just need to copy the >>module over, and you can leave the kernel alone. >> >>For developing a new driver, I really like having it as a module, >>cause you can load and unload and if it really screws up you just >>reboot. >> >>-- > > I have to agree with you, when developing it's useful, if it doesn't work, > unload and upload a new version. But it's about the framebuffer, it just > doesn't show up... > But I recall now that after compiling the framebuffer as a part of the > kernel, I have a file "fb" in my /proc folder... may it be the Graal I'm > looking for ?? > Then how do I access it from the kernel driver ? > > Thanks for your answer, > Julien > > _________________________________________________________________ > MSN Messenger : discutez en direct avec vos amis ! > http://www.msn.fr/msger/default.asp > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share > your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: Chris D. <chr...@gm...> - 2007-02-23 22:23:53
Attachments:
kernel-2.6.18.config
|
I just udated to buildroot version r1303 and I'm getting the same error when the kernel tries to load the pxafb driver: pxa2xx-fb: probe of pxa2xx-fb failed with error -12 It looks like there was a big change in the pxafb kernel driver somewhere between 2.6.18 and 2.6.20, and I suspect that this error is somehow related to those changes. I have a 240x320 passive lcd hooked up to my breakout-gs and I can tell you that it does work with a 2.6.18 kernel. The change to 2.6.20 came with r1286, so if you checkout a revision before that you'll get an older kernel. I'm attaching the kernel config that I used for my 2.6.18 kernel. It has support for the console on the framebuffer display, which enables you to see the boot messages. It also has some other stuff turned on/off so you might just want to use it as a rough guide. For bootargs I use: bootargs=console=tty0 console=ttyS0,115200n8 root=1f01 rootfstype=jffs2 reboot=cold,hard video=pxafb:mode:240x320-16,passive,hsynclen:2,right:1,left:1,upper:0,lower:0,vsynclen:3,hsync:1,vsync:1,pixclockpol:0,pixclock:300000 fbcon=rotate:1 and this is what dmesg shows for the pxafb driver: pxa2xx-fb pxa2xx-fb: overriding resolution: 240x320 pxa2xx-fb pxa2xx-fb: overriding bit depth: 16 pxa2xx-fb pxa2xx-fb: override hsynclen: 2 pxa2xx-fb pxa2xx-fb: override right: 1 pxa2xx-fb pxa2xx-fb: override left: 1 pxa2xx-fb pxa2xx-fb: override upper: 0 pxa2xx-fb pxa2xx-fb: override lower: 0 pxa2xx-fb pxa2xx-fb: override vsynclen: 3 pxa2xx-fb pxa2xx-fb: override hsync: Active High pxa2xx-fb pxa2xx-fb: override vsync: Active High pxa2xx-fb pxa2xx-fb: override pixel clock polarity: falling edge pxa2xx-fb pxa2xx-fb: override pixclock: 300000 Console: switching to colour frame buffer device 40x30 and then you get the /dev/fb0 entry Hope this helps. Chris On 2/23/07, Alexis Chiarello <chi...@ya...> wrote: > Julien, > > Any progress on the framebuffer side ?? > > I think I have the same issues... Framebuffer driver is built in the kernel > : > > # > # Graphics support > # > # CONFIG_FIRMWARE_EDID is not set > CONFIG_FB=y > CONFIG_FB_CFB_FILLRECT=y > CONFIG_FB_CFB_COPYAREA=y > CONFIG_FB_CFB_IMAGEBLIT=y > # CONFIG_FB_MACMODES is not set > # CONFIG_FB_BACKLIGHT is not set > CONFIG_FB_MODE_HELPERS=y > # CONFIG_FB_TILEBLITTING is not set > # CONFIG_FB_S1D13XXX is not set > CONFIG_FB_PXA=y > CONFIG_FB_PXA_PARAMETERS=y > # CONFIG_FB_MBX is not set > # CONFIG_FB_VIRTUAL is not set > > While it is booting I get this : > > pxa2xx-fb: probe of pxa2xx-fb failed with error -12 > > There is nothing before or after in the log that can help, but you can find > the full bootlog attached. > > I am using distrib release 1295 and my config is a basix 400 XM with > breakout GS. BTW, nothing yet is connected to the LCD pins on the > breakout-GS, but I don't think this is the issue. > > Of course, I cannot see the /dev/fb0, I only get the /proc/fb. > > I tried as a module, and it gives me the same error. > > Any help would be appreciated, and I don't seem to be the only one in that > situation. > > Best, > > Alexis. > ----- Original Message ----- > From: "Julien Lebot" <les...@ho...> > To: <gum...@li...> > Sent: Friday, February 23, 2007 1:47 AM > Subject: Re: [Gumstix-users] About the gumstix kernel / UCB1400 / > Framebuffer > > > > Hi Dave, > > > > That's me, Julien, who posted the question, just to make sure nobody gets > > confused, I'm not talking for Steve :) > > > >>Hi Steve, > >> > >> > Ok now a couple of nasty question (sorry, noob inside): > >> > > >> > First, can I transfer only the kernel to the gumstix without having to > >>flash > >> > it ? Because I don't touch anymore the package selection, just the > >>kernel > >> > configuration. > >> > >>Provided that the kernel that was there came from your buildroot, and > >>you have enough space (preferably an XM), then yeah you can do it. > >> > >>You could remove the old uImage and copy the new one in. Worst case, > >>is you need to download the whole thing. > >> > >>If you compiled the driver as a module, then you just need to copy the > >>module over, and you can leave the kernel alone. > >> > >>For developing a new driver, I really like having it as a module, > >>cause you can load and unload and if it really screws up you just > >>reboot. > >> > >>-- > > > > I have to agree with you, when developing it's useful, if it doesn't work, > > unload and upload a new version. But it's about the framebuffer, it just > > doesn't show up... > > But I recall now that after compiling the framebuffer as a part of the > > kernel, I have a file "fb" in my /proc folder... may it be the Graal I'm > > looking for ?? > > Then how do I access it from the kernel driver ? > > > > Thanks for your answer, > > Julien > > > > _________________________________________________________________ > > MSN Messenger : discutez en direct avec vos amis ! > > http://www.msn.fr/msger/default.asp > > > > > > ------------------------------------------------------------------------- > > Take Surveys. Earn Cash. Influence the Future of IT > > Join SourceForge.net's Techsay panel and you'll get the chance to share > > your > > opinions on IT & business topics through brief surveys-and earn cash > > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > > _______________________________________________ > > gumstix-users mailing list > > gum...@li... > > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > > |
From: Julien L. <les...@ho...> - 2007-02-24 09:47:33
|
Hi Alexis, >Julien, > >Any progress on the framebuffer side ?? Not really, according to another mailing list, if you can't read anything (even if you echo things in it) from the proc/fb it means it's compiled but not functional. >I think I have the same issues... Framebuffer driver is built in the kernel >: > ># ># Graphics support ># ># CONFIG_FIRMWARE_EDID is not set >CONFIG_FB=y >CONFIG_FB_CFB_FILLRECT=y >CONFIG_FB_CFB_COPYAREA=y >CONFIG_FB_CFB_IMAGEBLIT=y ># CONFIG_FB_MACMODES is not set ># CONFIG_FB_BACKLIGHT is not set >CONFIG_FB_MODE_HELPERS=y ># CONFIG_FB_TILEBLITTING is not set ># CONFIG_FB_S1D13XXX is not set >CONFIG_FB_PXA=y >CONFIG_FB_PXA_PARAMETERS=y ># CONFIG_FB_MBX is not set ># CONFIG_FB_VIRTUAL is not set > >While it is booting I get this : > >pxa2xx-fb: probe of pxa2xx-fb failed with error -12 > >There is nothing before or after in the log that can help, but you can find >the full bootlog attached. > >I am using distrib release 1295 and my config is a basix 400 XM with >breakout GS. BTW, nothing yet is connected to the LCD pins on the >breakout-GS, but I don't think this is the issue. > >Of course, I cannot see the /dev/fb0, I only get the /proc/fb. > >I tried as a module, and it gives me the same error. > >Any help would be appreciated, and I don't seem to be the only one in that >situation. > >Best, > >Alexis. Yep we have the same version (though I've tried early version too), and it fails with the -12 error... I see that you have the same BUG: at kernel/sched.c:4035 __schedule() [<c0021d60>] (dump_stack+0x0/0x14) from [<c0189190>] (__schedule+0x6f0/0x740) [<c0188aa0>] (__schedule+0x0/0x740) from [<c01893d4>] (schedule+0xd0/0x118) [<c0189304>] (schedule+0x0/0x118) from [<c00423d0>] (kthread+0xc8/0x128) r4 = 00000000 [<c0042308>] (kthread+0x0/0x128) from [<c0030330>] (do_exit+0x0/0x85c) r7 = 00000000 r6 = 00000000 r5 = 00000000 r4 = 00000000 bug... may it be that. In fact I'm trying to rewrite the pxafb driver for my needs, because I don't need the built in LCD controller, just the framebuffer, and it looks like the schedule part may be faulty (from this error). However I'm no way specialist in that. As I said I don't need the pxa controller, just the framebuffer, my LCD doesn't have any hsync or pixel clock stuff, so it needs a particular driver that I wrote... I need the framebuffer so my driver will read periodically from it and display its content on the screen... (that's the first step) I have a PSP LCD screen waiting to be wired but I keep it for when everything will be working (second step :) ) I'll keep you informed if I achieve something from my side. regards, Julien _________________________________________________________________ MSN Messenger : appels gratuits de PC à PC ! http://www.msn.fr/newhotmail/Default.asp?Ath=f |
From: Alexis C. <chi...@ya...> - 2007-02-23 23:11:08
|
Thanks Chris, it surely helps. I guess I have no choice going back to 2.6.18, but that's actually not a problem. Best, Alexis. ----- Original Message ----- From: "Chris Dollar" <chr...@gm...> To: "General mailing list for gumstix users." <gum...@li...> Sent: Friday, February 23, 2007 5:23 PM Subject: Re: [Gumstix-users] About the gumstix kernel / UCB1400 / Framebuffer >I just udated to buildroot version r1303 and I'm getting the same > error when the kernel tries to load the pxafb driver: > pxa2xx-fb: probe of pxa2xx-fb failed with error -12 > It looks like there was a big change in the pxafb kernel driver > somewhere between 2.6.18 and 2.6.20, and I suspect that this error is > somehow related to those changes. > > I have a 240x320 passive lcd hooked up to my breakout-gs and I can > tell you that it does work with a 2.6.18 kernel. The change to 2.6.20 > came with r1286, so if you checkout a revision before that you'll get > an older kernel. I'm attaching the kernel config that I used for my > 2.6.18 kernel. It has support for the console on the framebuffer > display, which enables you to see the boot messages. It also has some > other stuff turned on/off so you might just want to use it as a rough > guide. > > For bootargs I use: > bootargs=console=tty0 console=ttyS0,115200n8 root=1f01 > rootfstype=jffs2 reboot=cold,hard > video=pxafb:mode:240x320-16,passive,hsynclen:2,right:1,left:1,upper:0,lower:0,vsynclen:3,hsync:1,vsync:1,pixclockpol:0,pixclock:300000 > fbcon=rotate:1 > > and this is what dmesg shows for the pxafb driver: > pxa2xx-fb pxa2xx-fb: overriding resolution: 240x320 > pxa2xx-fb pxa2xx-fb: overriding bit depth: 16 > pxa2xx-fb pxa2xx-fb: override hsynclen: 2 > pxa2xx-fb pxa2xx-fb: override right: 1 > pxa2xx-fb pxa2xx-fb: override left: 1 > pxa2xx-fb pxa2xx-fb: override upper: 0 > pxa2xx-fb pxa2xx-fb: override lower: 0 > pxa2xx-fb pxa2xx-fb: override vsynclen: 3 > pxa2xx-fb pxa2xx-fb: override hsync: Active High > pxa2xx-fb pxa2xx-fb: override vsync: Active High > pxa2xx-fb pxa2xx-fb: override pixel clock polarity: falling edge > pxa2xx-fb pxa2xx-fb: override pixclock: 300000 > Console: switching to colour frame buffer device 40x30 > > and then you get the /dev/fb0 entry > > Hope this helps. > > Chris > On 2/23/07, Alexis Chiarello <chi...@ya...> wrote: >> Julien, >> >> Any progress on the framebuffer side ?? >> >> I think I have the same issues... Framebuffer driver is built in the >> kernel >> : >> >> # >> # Graphics support >> # >> # CONFIG_FIRMWARE_EDID is not set >> CONFIG_FB=y >> CONFIG_FB_CFB_FILLRECT=y >> CONFIG_FB_CFB_COPYAREA=y >> CONFIG_FB_CFB_IMAGEBLIT=y >> # CONFIG_FB_MACMODES is not set >> # CONFIG_FB_BACKLIGHT is not set >> CONFIG_FB_MODE_HELPERS=y >> # CONFIG_FB_TILEBLITTING is not set >> # CONFIG_FB_S1D13XXX is not set >> CONFIG_FB_PXA=y >> CONFIG_FB_PXA_PARAMETERS=y >> # CONFIG_FB_MBX is not set >> # CONFIG_FB_VIRTUAL is not set >> >> While it is booting I get this : >> >> pxa2xx-fb: probe of pxa2xx-fb failed with error -12 >> >> There is nothing before or after in the log that can help, but you can >> find >> the full bootlog attached. >> >> I am using distrib release 1295 and my config is a basix 400 XM with >> breakout GS. BTW, nothing yet is connected to the LCD pins on the >> breakout-GS, but I don't think this is the issue. >> >> Of course, I cannot see the /dev/fb0, I only get the /proc/fb. >> >> I tried as a module, and it gives me the same error. >> >> Any help would be appreciated, and I don't seem to be the only one in >> that >> situation. >> >> Best, >> >> Alexis. >> ----- Original Message ----- >> From: "Julien Lebot" <les...@ho...> >> To: <gum...@li...> >> Sent: Friday, February 23, 2007 1:47 AM >> Subject: Re: [Gumstix-users] About the gumstix kernel / UCB1400 / >> Framebuffer >> >> >> > Hi Dave, >> > >> > That's me, Julien, who posted the question, just to make sure nobody >> > gets >> > confused, I'm not talking for Steve :) >> > >> >>Hi Steve, >> >> >> >> > Ok now a couple of nasty question (sorry, noob inside): >> >> > >> >> > First, can I transfer only the kernel to the gumstix without having >> >> > to >> >>flash >> >> > it ? Because I don't touch anymore the package selection, just the >> >>kernel >> >> > configuration. >> >> >> >>Provided that the kernel that was there came from your buildroot, and >> >>you have enough space (preferably an XM), then yeah you can do it. >> >> >> >>You could remove the old uImage and copy the new one in. Worst case, >> >>is you need to download the whole thing. >> >> >> >>If you compiled the driver as a module, then you just need to copy the >> >>module over, and you can leave the kernel alone. >> >> >> >>For developing a new driver, I really like having it as a module, >> >>cause you can load and unload and if it really screws up you just >> >>reboot. >> >> >> >>-- >> > >> > I have to agree with you, when developing it's useful, if it doesn't >> > work, >> > unload and upload a new version. But it's about the framebuffer, it >> > just >> > doesn't show up... >> > But I recall now that after compiling the framebuffer as a part of the >> > kernel, I have a file "fb" in my /proc folder... may it be the Graal >> > I'm >> > looking for ?? >> > Then how do I access it from the kernel driver ? >> > >> > Thanks for your answer, >> > Julien >> > >> > _________________________________________________________________ >> > MSN Messenger : discutez en direct avec vos amis ! >> > http://www.msn.fr/msger/default.asp >> > >> > >> > ------------------------------------------------------------------------- >> > Take Surveys. Earn Cash. Influence the Future of IT >> > Join SourceForge.net's Techsay panel and you'll get the chance to share >> > your >> > opinions on IT & business topics through brief surveys-and earn cash >> > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV >> > _______________________________________________ >> > gumstix-users mailing list >> > gum...@li... >> > https://lists.sourceforge.net/lists/listinfo/gumstix-users >> >> ------------------------------------------------------------------------- >> Take Surveys. Earn Cash. Influence the Future of IT >> Join SourceForge.net's Techsay panel and you'll get the chance to share >> your >> opinions on IT & business topics through brief surveys-and earn cash >> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV >> _______________________________________________ >> gumstix-users mailing list >> gum...@li... >> https://lists.sourceforge.net/lists/listinfo/gumstix-users >> >> >> > -------------------------------------------------------------------------------- > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share > your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV -------------------------------------------------------------------------------- > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > |
From: Steven A. F. <sa...@op...> - 2007-02-24 03:25:31
|
I agree that modules can be useful. But what I found was that the frame-buffer module could not be unloaded - it always seemed to be busy. So I had to reboot anyway. And once I had to reboot, then it was easier to just compile the driver in. That is why I made my recommendation to compile it in - it eliminates one variable. Do you see any evidence that the driver is doing anything? Are there any messages when the kernel boots, or anything in dmesg? If there are no messages at all, then that suggests that the driver is not even being called. My next step would be to add a few printk() statements, to see what is happening. For example, the routines pxafb_init() and pxafb_probe() should be called, so adding printk there will tell you if the initialization is happening. If you have never used printk, it works just like printf in user programs. Just give it a format string and arguments, and you should see messages on the console serial port. Simple usage would be just something like: printk("got here\n"); I used this technique when debugging my display. It is a little painful at first, but it will tell you whether the driver is even being called. Steve Julien Lebot wrote: > Hi Dave, > > That's me, Julien, who posted the question, just to make sure nobody gets > confused, I'm not talking for Steve :) > > >> Hi Steve, >> >> >>> Ok now a couple of nasty question (sorry, noob inside): >>> >>> First, can I transfer only the kernel to the gumstix without having to >>> >> flash >> >>> it ? Because I don't touch anymore the package selection, just the >>> >> kernel >> >>> configuration. >>> >> Provided that the kernel that was there came from your buildroot, and >> you have enough space (preferably an XM), then yeah you can do it. >> >> You could remove the old uImage and copy the new one in. Worst case, >> is you need to download the whole thing. >> >> If you compiled the driver as a module, then you just need to copy the >> module over, and you can leave the kernel alone. >> >> For developing a new driver, I really like having it as a module, >> cause you can load and unload and if it really screws up you just >> reboot. >> >> -- >> > > I have to agree with you, when developing it's useful, if it doesn't work, > unload and upload a new version. But it's about the framebuffer, it just > doesn't show up... > But I recall now that after compiling the framebuffer as a part of the > kernel, I have a file "fb" in my /proc folder... may it be the Graal I'm > looking for ?? > Then how do I access it from the kernel driver ? > > Thanks for your answer, > Julien > > _________________________________________________________________ > MSN Messenger : discutez en direct avec vos amis ! > http://www.msn.fr/msger/default.asp > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > |
From: Chris D. <chr...@gm...> - 2007-02-24 03:44:48
|
Hi Steve, Just curious if you've had a chance to try the framebuffer on the latest 2.6.20 kernel? I've been using the pxafb stuff since probably 2.6.13 with no issues till 2.6.20. (I've always compiled it directly into the kernel as well). Chris On 2/23/07, Steven A. Falco <sa...@op...> wrote: > > I agree that modules can be useful. But what I found was that the > frame-buffer module could not be unloaded - it always seemed to be busy. So > I had to reboot anyway. And once I had to reboot, then it was easier to > just compile the driver in. That is why I made my recommendation to compile > it in - it eliminates one variable. > > Do you see any evidence that the driver is doing anything? Are there any > messages when the kernel boots, or anything in dmesg? If there are no > messages at all, then that suggests that the driver is not even being > called. > > My next step would be to add a few printk() statements, to see what is > happening. For example, the routines pxafb_init() and pxafb_probe() should > be called, so adding printk there will tell you if the initialization is > happening. If you have never used printk, it works just like printf in user > programs. Just give it a format string and arguments, and you should see > messages on the console serial port. Simple usage would be just something > like: > > printk("got here\n"); > > I used this technique when debugging my display. It is a little painful at > first, but it will tell you whether the driver is even being called. > > Steve > > > > Julien Lebot wrote: > Hi Dave, > > That's me, Julien, who posted the question, just to make sure nobody gets > confused, I'm not talking for Steve :) > > > > Hi Steve, > > > > Ok now a couple of nasty question (sorry, noob inside): > > First, can I transfer only the kernel to the gumstix without having to > > flash > > > it ? Because I don't touch anymore the package selection, just the > > kernel > > > configuration. > > Provided that the kernel that was there came from your buildroot, and > you have enough space (preferably an XM), then yeah you can do it. > > You could remove the old uImage and copy the new one in. Worst case, > is you need to download the whole thing. > > If you compiled the driver as a module, then you just need to copy the > module over, and you can leave the kernel alone. > > For developing a new driver, I really like having it as a module, > cause you can load and unload and if it really screws up you just > reboot. > > -- > > I have to agree with you, when developing it's useful, if it doesn't work, > unload and upload a new version. But it's about the framebuffer, it just > doesn't show up... > But I recall now that after compiling the framebuffer as a part of the > kernel, I have a file "fb" in my /proc folder... may it be the Graal I'm > looking for ?? > Then how do I access it from the kernel driver ? > > Thanks for your answer, > Julien > > _________________________________________________________________ > MSN Messenger : discutez en direct avec vos amis ! > http://www.msn.fr/msger/default.asp > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > |
From: Julien L. <les...@ho...> - 2007-02-24 10:06:34
|
Hi Chris, > >Just curious if you've had a chance to try the framebuffer on the >latest 2.6.20 kernel? I've been using the pxafb stuff since probably >2.6.13 with no issues till 2.6.20. (I've always compiled it directly >into the kernel as well). > >Chris Looks like we'll have to downgrade to an earlier version to make the graphics / audio stuff working, that's paradoxal :p Julien _________________________________________________________________ Testez Windows Live Mail Beta ! http://www.ideas.live.com/ |