Hacking [Pre-release, WIP] Yet another rxTools?

[Madridi]

Spider works for me, even tried with additional forced reinit. But maybe not inside the error handler, since most probably there is no underruns with my card. Yes, please recompile and try once again, maybe it will help

It did work fine after the reinstall as mentioned in the edit. But still got a freeze with the emunand dump

 

[duke_srg]

It did work fine after the reinstall as mentioned in the edit. But still got a freeze with the emunand dump

So it looks like that Sd car itself must be reset somehow, not only controller. I'll search for the solution but that's all for today

 

[duke_srg]

Aded minor fixes, most probably won't fox the issue, but anyway it is worth to try

 

[Madridi]

Aded minor fixes, most probably won't fox the issue, but anyway it is worth to try

I'll test it later tonight. I'm out almost all day long. Hope it fixes things finally

 

[Madridi]

Aded minor fixes, most probably won't fox the issue, but anyway it is worth to try

Sorry for the delay.
Just tried this, and well, it gets stuck on a black screen with a blank progress bar (initial setup)
I checked the sd afterwards to see if anything got processed, but not even the data folder was created

 

[duke_srg]

Sorry for the delay.
Just tried this, and well, it gets stuck on a black screen with a blank progress bar (initial setup)
I checked the sd afterwards to see if anything got processed, but not even the data folder was created

Have you tried with /date folder contents existing?

 

[Madridi]

Have you tried with /date folder contents existing?

Wanted to do so, but I guess I must've deleted it off my PC. I guess I could run an earlier commit to get it again. Worth trying?

 

[duke_srg]

Wanted to do so, but I guess I must've deleted it off my PC. I guess I could run an earlier commit to get it again. Worth trying?

Yes, sure! Just to make sure it boots or not

 

[Madridi]

Yes, sure! Just to make sure it boots or not

It doesn't boot. Black screen

 

[duke_srg]

It doesn't boot. Black screen

Got it. So most probably it readw with errors all the way

 

[duke_srg]

Added possible workaround, please check with your card.

 

[Madridi]

Added possible workaround, please check with your card.

It booted this time, and the dump completed
It seemed slow though. I didn't time it, I'll run it a second time to time it

 

[duke_srg]

It booted this time, and the dump completed
It seemed slow though. I didn't time it, I'll run it a second time to time it

Must be at maximum speed. Though if error handling routines activated, it could slowdown, yes.
You could check for sure commenting
if (error) {
...
}
operator in two places inside tmio.c

Thanks to nocash123
https://gbatemp.net/threads/dsi-reverse-engineering-sd-mmc-sdio-registers.395787/page-2#post-7104995

 

[Madridi]

Must be at maximum speed. Though if error handling routines activated, it could slowdown, yes.
You could check for sure commenting
if (error) {
...
}
operator in two places inside tmio.c

Thanks to nocash123

Well, no.. It froze again on my second dump.
Again, it's happening around the same place it happens everytime. I really feel there is something to it. Cant be a coincidence!

 

[duke_srg]

Well, no.. It froze again on my second dump.
Again, it's happening around the same place it happens everytime. I really feel there is something to it. Cant be a coincidence!

There are still several options.
I can made a custom built do display CID/CSD registers so we make sure the card is genuine SanDisk and not a fraud. Thought this won't be of much help, since in looks like functioning well in all other cases.
The other option is to backup, reformat the card with SD formatter in FULL/Overwrite mode and restore. The actual delay may be caused buy several unstable blocks which is still not replaced by the reserved sectors. Full reformat will force wear leveling to adjust those blocks.

 

[Madridi]

There are still several options.
I can made a custom built do display CID/CSD registers so we make sure the card is genuine SanDisk and not a fraud. Thought this won't be of much help, since in looks like functioning well in all other cases.
The other option is to backup, reformat the card with SD formatter in FULL/Overwrite mode and restore. The actual delay may be caused buy several unstable blocks which is still not replaced by the reserved sectors. Full reformat will force wear leveling to adjust those blocks.

Well, I doubt my card is fake. Haven't notices any weird behavior with it, and the retailer I bought it from was amazon (not a reseller, amazon themselves).. I'd be more inclined to think that it's something with my setup instead

As for the other option, don't have the time or hdd space for a back up currently, so that's gonna be a problem atm

Also, like you said, it works fine in other cases (particularly when div is div_4).. So I don't know

 

[duke_srg]

As for the other option, don't have the time or hdd space for a back up currently

There is one easier option to refresh the card contents.
Backup/restore emunand, then copy over all rxTools release files, then copy some junk files to occupy all free space of the card and remove them
If didn't help, you can also the copy over the rest of the card contents by smaller bulks.

 

[Madridi]

There is one easier option to refresh the card contents.
Backup/restore emunand, then copy over all rxTools release files, then copy some junk files to occupy all free space of the card and remove them
If didn't help, you can also the copy over the rest of the card contents by smaller bulks.

That still will take a long time, which I don't have. Also, after backing up and restoring emunand, what about the rest of my files?

 

[duke_srg]

That still will take a long time, which I don't have. Also, after backing up and restoring emunand, what about the rest of my files?

You just need to make backup and restore the emunand with emunand tool on the PC, that won't affect nothing and take several minutes.
Also filling the rest the card free space with, say, copies of emunand.bin will be also quick.
Both will eliminate possible emunand and FAT partition problematic sectors which could cause the delays and underruns.
As you said it is always happens in approximately the same place, so this is possible. With the lower speed it still could be read with no issues, when emuNAND booting and working those sectors could be rarely used, and when accessing from the card reader, all underruns are handled properly, so you don't see any issues.
I know, we must handle those too in the rxTools, but no solution for now.

 

[Madridi]

You just need to make backup and restore the emunand with emunand tool on the PC, that won't affect nothing and take several minutes.
Also filling the rest the card free space with, say, copies of emunand.bin will be also quick.
Both will eliminate possible emunand and FAT partition problematic sectors which could cause the delays and underruns.
As you said it is always happens in approximately the same place, so this is possible. With the lower speed it still could be read with no issues, when emuNAND booting and working those sectors could be rarely used, and when accessing from the card reader, all underruns are handled properly, so you don't see any issues.
I know, we must handle those too in the rxTools, but no solution for now.

Emunand tool? I thought you meant backups from d9 or something. Emunand tool according to several devs is outdated now and might be dangerous to use. I'm not comfortable using that as I don't really want to lose anything.

If d9 still is fine to use for this experiment, then ok, I'll do the backup and restore later tonight, then fill the card up. After filling it up, do I just delete what I filled, copy rxTools to sd again, and try?

Yeah, it's still a matter that has to be within rxtools no matter what the result is. In which case, I suppose div_4 should be the current solution, even if it gains no speed improvements. Until a solution is found that is.