betacentauri
commented
2 years ago Does it work if you first stop e2 and then execute ofgwrite via telnet (complete command see above in the log).
Open technic opened 2 years ago
betacentauri
commented
2 years ago Does it work if you first stop e2 and then execute ofgwrite via telnet (complete command see above in the log).
technic
commented
2 years ago Does it work if you first stop e2 and then execute ofgwrite via telnet (complete command see above in the log).
yes
betacentauri
commented
2 years ago I know that some boxes prevent opening the frame buffer twice. Therefore I close and open the frame buffer again after e2 has stopped.
https://github.com/oe-alliance/ofgwrite/blob/master/ofgwrite.c#L854
So at the moment I don’t have an idea how to fix that. Maybe wait a little bit more?
betacentauri
commented
2 years ago Maybe add some more debug messages. Are you able to do that? Or should I offer a binary? (May take some days)
technic
commented
2 years ago thanks for your answers. I can take a closer look at this during the weekends. Interestingly, this issue does not occur 100% of the time, but I didn't find the reason yet. I don't have a toolchain at hand now, but I can compile it, it will be easier to experiment then. If you have ideas where to put more printfs, send me a binary or a patch.
betacentauri
commented
2 years ago I would try to add a “sleep(1)” or similar before opening the framebuffer. And also some debug messages before opening the framebuffer to exactly see which statement cause the problem and what the return value from e.g. open_framebuffer is. https://github.com/oe-alliance/ofgwrite/blob/master/fb.c#L586
If eg open_framebuffer is really successful and the driver afterwards “says” framebuffer is not open, then it would be a driver bug for me…
technic
commented
2 years ago Actually I don't see why it stops working when there is a framebuffer error, because the return code is ignored.
technic
commented
2 years ago So when there is a framebuffer not available. error the program crashes with SIGSEGV
...
user.warn kernel: [21:14:36:727 ERROR-HI_FB]:OPTM_GfxOpenLayer[2334]:you should initial and open display first, open fb0 failure!
user.warn kernel: [21:14:36:737 ERROR-HI_FB]:DRV_HIFB_Open[5579]:failed to open layer0 !
user.warn kernel: [21:14:36:950 ERROR-HI_FB]:DRV_HIFB_Open[5661]:hifb open failed
user.info ofgwrite: framebuffer not available.
user.info kernel: potentially unexpected fatal signal 11.
user.warn kernel: CPU: 0 PID: 1993 Comm: ofgwrite_bin Tainted: P C O 4.4.35 #1
user.warn kernel: Hardware name: bigfish
user.warn kernel: task: cf20d980 ti: cf3d4000 task.ti: cf3d4000
user.warn kernel: PC is at 0x1b270
user.warn kernel: LR is at 0x0
user.warn kernel: pc : [<0001b270>] lr : [<00000000>] psr: 800f0010
user.warn kernel: sp : be74d1b0 ip : 00000001 fp : 00000000
user.warn kernel: r10: 00000000 r9 : 000c1be8 r8 : 000be180
user.warn kernel: r7 : 00090bb3 r6 : 000c1b48 r5 : 00000000 r4 : b601e000
user.warn kernel: r3 : 000002d0 r2 : 00000500 r1 : 00000000 r0 : 00000000
user.warn kernel: Flags: Nzcv IRQs on FIQs on Mode USER_32 ISA ARM Segment user
user.warn kernel: Control: 10c53c7d Table: 10850059 DAC: 00000055
user.warn kernel: CPU: 0 PID: 1993 Comm: ofgwrite_bin Tainted: P C O 4.4.35 #1
user.warn kernel: Hardware name: bigfish
user.warn kernel: [<c001e830>] (unwind_backtrace) from [<c001b064>] (show_stack+0x20/0x24)
user.warn kernel: [<c001b064>] (show_stack) from [<c036cef0>] (dump_stack+0x24/0x28)
user.warn kernel: [<c036cef0>] (dump_stack) from [<c0017b50>] (show_regs+0x1c/0x20)
user.warn kernel: [<c0017b50>] (show_regs) from [<c0038c60>] (get_signal+0x594/0x598)
user.warn kernel: [<c0038c60>] (get_signal) from [<c001a6c8>] (do_signal+0xd0/0x410)
user.warn kernel: [<c001a6c8>] (do_signal) from [<c001abe4>] (do_work_pending+0xac/0xbc)
user.warn kernel: [<c001abe4>] (do_work_pending) from [<c00165b4>] So maybe there is subsequent read or write which goes wrong
technic
commented
2 years ago Ok, it is quite clear that mmap_fb did not succeed and all subsequent writes to fb will crash.
Maybe we should check init_framebuffer exit code?
betacentauri
commented
2 years ago And then? Stop ofgwrite or flash without showing the progress on the TV screen?
init_framebuffer is called 2 times. Can you see which call fails? I guess the second one.
Well, not very good coding from me :-( But I would like to see more debug messages to understand what goes wrong exactly. I'll prepare a patch for you...
betacentauri
commented
2 years ago I hope the binary works (is build for ARM boxes). If not, patch is included.
technic
commented
2 years ago And then?
Maybe go to reboot,... well, there is no easy way to report an error actually
I guess the second one.
I think it's this one https://github.com/oe-alliance/ofgwrite/blob/master/ofgwrite.c#L854 but I am not sure. Thanks for the binary. This issue popups not every time, I will try to get a new log.
betacentauri
commented
2 years ago Line 854: Yes, I think so, too.
The binary tries now to open the framebuffer 2 times with 5 seconds delay after first try. So even if it’s successful, please look into the log whether first try failed.
technic
commented
2 years ago I'll let you know later, it's not compatible with armv7
betacentauri
commented
2 years ago Was for a sf8008 as build env was ready for this box and I thought all hisi boxes use similar arm chip...
I'll try to build new binary...
betacentauri
commented
2 years ago ofgwrite_framebuffer_log.zip Now it should work. Replace the ofgwrite_bin in /usr/bin/ with the unzipped ofgwrite_bin.
betacentauri
commented
2 years ago Any feedback regarding this?
technic
commented
2 years ago Hi, Sorry, I got occupied with other things. Thanks for the reminder I hope I can check it after holidays.
Hi, I am trying to use online flash in openatv 6.4 image on axashistwin. After launching ofgwrite it goes until
init 2stage, and then I see black screen. After ssh to the box I see that no ofgwrite and no enigma2 process is running. According to enigma2 log I haveThere is the following message in the /var/log/messages: