-
-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Kobo Clara 2E] Koreader freezes randomly #9806
Comments
A few workarounds have been implemented in current nightlies, but the Clara 2E (like the Libra 2 & Nia) runs on a severely capricious board, so I can't guarantee a miracle cure ;). |
I see, I guess I'll be on the lookout for improvements after the next release. |
Not sure if this is related, but I have noticed Koreader freezing when I try to jump to a chapter using the TOC. What fixed that for me was deleting fontinfo.dat, but I have had to do that twice now because it kept happening. |
Wrong deduction ;). One is just the consequence of the other, but not this way around. That specific behavior is known to be extremely deadly to those capricious boards (#8414), which is why part of the workarounds involve disabling the racy bits of this specific interaction. And the consequence of a hang is a hard-reboot, which is a very fine way to trigger filesystem corruption. Unlike most of our settings, the |
Sadly this doesn't seem to be the case for me. I've been using the 20/11 nightly and this behavior persists, sadly these workarounds aren't too effective in my case (which you said might be the case :p). crash.log |
Logs are woefully unhelpful unless they're verbose debug logs ;). Keep in mind it's unlikely to be terribly helpful here anyway, since the hard-reboot will likely cause truncation (and since we're not the ones crashing, it wouldn't be terribly helpful anyway ^^). This would probably require a live serial console to catch the actual kernel BUG/OOPS/PANIC, and possibly a custom more verbose kernel build to actually figure it out (... and the Clara 2E kernel hasn't been released yet anyway). |
Just wanted to note this is not out of date. It still happens even with the latest firmware (22.11). |
Not so small nit: Don't confuse FW versions (i.e., Nickel, currently at 4.34 or 4.35), and KOReader releases (currently 2022.11) ;). |
Point taken! Thank you, I stand corrected.
…On Sun, Dec 25, 2022, 20:23 NiLuJe ***@***.***> wrote:
Not so small nit: Don't confuse FW versions (i.e., Nickel, currently at
4.34 or 4.35), and KOReader releases (currently 2022.11) ;).
—
Reply to this email directly, view it on GitHub
<#9806 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAIKXRADI26FDY34ENGWIXTWPCGKLANCNFSM6AAAAAASE3YIVY>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
I am here to thank you guys for helping users with KOReader. Since random freezes are a known issue, I would like to know if there is a possible solution in sight for this problems or KOReader experience is bound to be broken on Clara 2E forever. I am not an expert, so please forgive my ignorance. |
I also have the same problem. These freezes are happening randomly (during page turning, brightness adjustment, menu navigation, etc) once in an hour. |
From my side, I'm about to give up on the Clara 2E as a compatible device. I think if the issue is device-specific, the KOReader docs should probably be updated to recommend people don't run it on this device (KOReader was the main reason I bought this unit). |
I've gone ahead and edited the Kobo page on the Wiki to indicate that the Clara 2E is only partially supported -- hopefully this would save others an expense on a device that doesn't work right. |
@ezuk so, bad news in sight!? KOReader was the only reason why I bought it... Should I consider resell to someone who won't use KOReader? |
I think I am going to try and resell mine, yes... Same thing - KOReader was the only reason I got this device. |
Unless someone can get me, at the very least, the actual kernel log from the OOPS/PANIC, no, there won't be any traction on this. (Given that it seems like the watchdog gets tripped, you can't even hope to get it via USBNet after a hang, so this would probably require serial access). And it would also probably require a custom kernel with extra epdc debugging to actually investigate... and the Clara 2E kernel sources haven't been released by Kobo ;o). |
Would having an actual device help at all? I can just send you mine (you won't have to send it back). |
If the watchdog weren't an issue, I would have said yes, but I'm not a hardware tinkerer, so popping it open to get serial access is not going to happen (I have neither the tools nor the skills required), which means I wouldn't be able to do much with it :/. (Thanks, though!). |
I'm also not going to lie and add that there's a bit of a "this should be the OEM's job" (of shipping actually working stuff) going on here, which adds to me not really be willing to sink my already very limited time into this. |
I completely agree. The only thing giving me pause is that it doesn't seem to freeze on the native UI. Maybe they specced it very tightly and whenever we exceed something it breaks. It's hard for me to raise a warranty issue with the OEM because the base firmware seems to work. A tricky situation... That's why I think the best solution is to explicitly exclude this device from the list of supported devices. |
The Libra 2, which is based on a very similar board, exhibits similar
issues in Nickel, too.
If that's truly not the case on the Clara 2e, then this gets even more
mysterious...
…On Tue, Jan 10, 2023, 10:02 Erez Zukerman ***@***.***> wrote:
I'm also not going to lie and add that there's a bit of a "this should be
the OEM's job" (of shipping actually working stuff)
I completely agree. The only thing giving me pause is that it doesn't seem
to freeze on the native UI. Maybe they specced it very tightly and whenever
we exceed something it breaks. It's hard for me to raise a warranty issue
with the OEM because the base firmware seems to work. A tricky situation...
That's why I think the best solution is to explicitly exclude this device
from the list of supported devices.
—
Reply to this email directly, view it on GitHub
<#9806 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAA3KZU6TVCMGIX5FI2OASTWRUQTJANCNFSM6AAAAAASE3YIVY>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
I assume this means you either ported your config from another device where you had it tweaked, or you tweaked it at one point. Remove the (I'd like to confirm the automatic logic behaves, which is why I'm not saying: yeah, just tick it ;)). |
There is no entry in
|
Ohkay, I'll double-check the logic then, thanks ;o). |
Nope, I definitely can't reproduce that :/. But a quick look at your (Because it's from #9691 for the Clara 2E, which post-dates that). |
Oh sorry about that. I never updated past my intial install because I have some custom scripts I want to keep. I saw #7431 was in 2021 so that's why I presumed I was up-to-date enough. Still super weird that an active SSH session prevents the issue. BTW: thanks @ezuk for your heads-up that the SD is remove-able, good to know. |
On which version?
No, but it affects the syslog: check
Eeeh, jury's still out on that. It's a different symptom, and more benign, buuuuuuut....
Which issue? The full hang => watchdog reboot stuff is magically prevented by being plugged in or having Wi-Fi enabled (probably because power), so, since SSH involves either or those, not too surprising ;). |
The kernel source has been released on github/kobolabs. Not sure if it is complete source or not. |
for this bug, it seems from low level (hardware IO) and need to be handled inside kernel. This is the source: https://community.nxp.com/t5/i-MX-Processors-Knowledge-Base/uSDHC-auto-tuning-and-possible-SDIO-failures/ta-p/1352855 also maybe related: https://community.nxp.com/t5/i-MX-Processors/Why-does-my-i-MX6UL-based-Linux-system-freezes-when-accessing-IO/m-p/901864#M136043 |
I'm aware ;). |
@ezuk kindly provided a device so that I may suffer with all of you ;p. The amount of time the watchdog (?) takes to reboot the device is... weirdly fluid. My first attempt at getting it to hang took some mighty effort (by basically spending a couple minutes squiggling with finger_trace), and that took a very long while to reboot. Another extremely confusing thing: I have a significantly harder time getting it to hang if I... simply enable klogd & swap to a "to disk" syslogd. I haven't isolated if only klogd "helps" here, but that's super weird. (Also, I bump the console logging level to max via (Note that this doesn't really definitely help, as that's the state I managed to hang it in via finger_trace). TL;DR: Mysterious mystery is mysterious. |
Spoiler alert: it doesn't. |
i.e., the last thing logged by the kernel is a possibly not actually relevant (i.e., stale)
Which I'm pretty sure you don't actually need elevated logging to see, because that typo is familiar to me ;). (And makes perfect sense in the context of ftrace, so, it's not spurious at all, it's the driver doing its job). |
Sidebar: but not with getting some wait_for_completion ioctls to timeout, though. That happened fairly quickly regardless. (I've disabled all the current workarounds for these tests, of course). |
Great that you got your hands on a device! That's very nice from @ezuk . Hopefully you can work your magic and figure this mysterious mystery out. |
Just so I have stuff to test, any specific patterns you guys remember as "crash-prone"? I haven't managed to get a Notification to hang so far, for example, but I've only tested basic stuff with a single notification (mainly frontlight stuff). So far, my go-to is the jump marker, as that seems to be fairly "reliable" (when the kernel isn't being weird, at least ;p). |
Yes the only freezes I have experienced are seemingly random ones, and after following in document links, like from a TOC, or footnotes etc. |
Well, since disabling the notifications I've been having a very smooth experience. Ignoring a couple of weeks during which the inner sd card got screwy and my device kept hanging (not crashing), got that solved by replacing it. |
I keep thinking that the "close with comment" is an option I need to disable before I press "comment" on my phone >:| |
Another fun fact: someone must be aware that their Elan panel implementation has terrible accuracy issues (or, more specifically, properly registering stuff at all), because, when you enable Nickel's developer mode, the power LED will light up on registered touches ;p. (Which I imagine serves as another reminder that "yup, we're not the ones fucking up, that kernel just sucks ass") ;).) |
…XP boards (#10771) * Notification: Drop the fencing from #10083; it never actually helped, and had subtle side-effects we could do without. * VirtualKeyBoard: Flash on close, otherwise, some of the fast refresh glitches may be burned into the working buffer until a flash. Making sure we flash ourselves prevent it from sticking around on the page ;). * util: Move `writeToSysfs` to base (i.e., `ffi/util`), as we need it there (and it actually makes more sense there anyway ;p). * Bump base for koreader/koreader-base#1645, which is where the actual workaround (hopefully) lives. Re #8414, #9806, #10558
After disabling "Battery level" in "Device status alert" no more freezes for me |
On a positive note; I've had the Clara 2E since last november, installed the 2022.11 release and didnt have any problems. Its just that I'm a bit afraid to "shake the jelly" when I update. Should I keep this combination of hard&software, or would a new release most probably not regress regarding this hw? |
The aforementioned workarounds landed in 2023.08, make of that what you will ;). |
I've ve owned the libra 2 (latest firmare and koreader version installed) for a few weeks now. Been reading over 5 books, not a single issue so far. Either I'm lucky or you guys did a great job :) |
(Via #10771) |
Issue
The device randomly freezes, usually while reading but it happened at least once while browsing the settings. Doesn't happen often, about once a day, forcing me to shutdown the device using the power button.
Steps to reproduce
Use koreader for several hours.
crash.log
(if applicable)crash.log
Btw
Thanks for this piece of software, it works great most of the time :D
The text was updated successfully, but these errors were encountered: