Trying to play tracks crashes with current dev build on iPod mini 2nd Gen

[iPodVT]

This evening I used Rockbox Utility to update my iPod mini 2nd Gen's installation of dev build a7cfee640d-240908 to the current dev build 2c1d31fcd9-250210.  After rebooting into the new dev build I was unable to play any tracks - neither via the Database nor via the File Browser.  On my first try with via Database the iPod froze so I hard rebooted (Menu+Select) and tried the File Browser and that gave me this crash screen:

Data abort at 0001ca6c
pc: 0001ca6c
sp: 4000b660
    A:  00031168
    A:  0002e130
    A:  00017150
    A:  00017db8
bt end

So then I reverted to my backup of a7cfee640d-240908 and was able to play tracks as normal via both the Database and the File Browser.

Then I used Rockbox Utility again to do a clean install of 2c1d31fcd9-250210 instead of updating a7cfee640d-240908.  Rather than rebuilding the Database, I just tried running from the File Browser.  This time the crash screen displayed a slight variation from the previous one above:

Data abort at 0001ca74
pc: 0001ca74
sp: 4000b660
    A:  00031168
    A:  0002e130
    A:  00017150
    A:  00017db8
bt end

Then I rebooted and tried building the database for the clean-installed dev build, and the iPod froze.

This iPod has an iFlash SD-CF with a 256GB Samsung Pro Plus microSD in an SD adapter.

[speachy]

This evening I used Rockbox Utility to update my iPod mini 2nd Gen's installation of dev build a7cfee640d-240908 to the current dev build 2c1d31fcd9-250210.

A five-month spread is pretty wide.

I suggest you try something in the middle, like the oldest available daily build, and if that works, work forward until you find the date at which things go awry?

[iPodVT]

I suggest you try something in the middle, like the oldest available daily build, and if that works, work forward until you find the date at which things go awry?

I was afraid you might say that... ;-)

I worked my way up to dev build 95f4accf45, which played.  It's immediate successor, 2e98889be3, froze.

For good measure I tried 2c1d31fcd9-250210 on an ipod6g and it didn't freeze or crash, though I only tried it briefly to see if it would play at all.  I will continue to use that one on the 6G, and I'll run 95f4accf45 on the mini until there's a relevant new version posted on the Recent Activity page or an exp version gets posted to this topic.

Thanks.

[iPodVT]

I worked my way up to dev build 95f4accf45, which played.  It's immediate successor, 2e98889be3, froze.

Immediate successor among the daily builds, that is.

[speachy]

I worked my way up to dev build 95f4accf45, which played.  It's immediate successor, 2e98889be3, froze.

FWIW as I type this I'm listening to music on a mini2g running last night's build (1ea5d57303).

It's a selection of Opus, WMA, AAC, FLAC, and MP3 that I use for basic smoke tests.

Please give that build a whirl, and if it still goes boom try with a clean .rockbox directory before applying your settings/themes one by one.

[iPodVT]

Thanks a lot for your reply - I really appreciate it.

Earlier this morning, when I saw that the new 1ea5d57303 dev build had been posted and was reported to have fixed a crashing bug, I used Rockbox Utility to update from 95f4accf45 on my mini 2G with all of my configs etc.  It froze when I tried to play a track, so I reverted back to 95f4accf45 and I just now finished playing the entire album that includes that track.  But I should have tried a clean install of 1ea5d57303.  I have to go out now for a bit but I will try a clean install when I return, as well as anything else you suggest/request.  Meanwhile, there's no urgency on my part for this fix.

Again, thanks.

[iPodVT]

Now I have tried a clean install of 1ea5d57303 on the same iPod mini 2nd Gen and it froze again when I tried to play a track.  Then I tried a clean install of 1ea5d57303 on a different iPod mini 2nd Gen with a 4GB SanDisk CF Card and it also froze when I tried to play a track.  Both iPods had plenty of free/available disk space.

Please let me know if there is anything else you would like me to try.

Thanks.

[iPodVT]

FWIW as I type this I'm listening to music on a mini2g running last night's build (1ea5d57303).

It's a selection of Opus, WMA, AAC, FLAC, and MP3 that I use for basic smoke tests.

Does your mini2g have an HDD for storage?  If not, what flavor of flash?

I tried a mini2g with a mechanical HDD (albeit a 5GB HDD from a scavenged/canabilized Dell Pocket DJ) and had the same results as with the others:  runs fine on older dev build but freezes with 1ea5d57303 when trying to play or navigate the File Browser.

[iPodVT]

I tried a mini2g with a mechanical HDD (albeit a 5GB HDD from a scavenged/canabilized Dell Pocket DJ) and had the same results as with the others:  runs fine on older dev build but freezes with 1ea5d57303 when trying to play or navigate the File Browser.

That was with a clean install of 1ea5d57303.

[Bilgus]

There are several builds between those can you try the two others here?
https://www.mediafire.com/folder/mbwla8c3rjqea/ipodmini_2g_crash_builds

I also included the one you referenced as good and head incase speachy fixed it already

[iPodVT]

You just barely caught me - I was about to hit the hay - but I will give them a whirl and let you know before I turn in.

[iPodVT]

This one froze when I tried to play a track from the File Browser:
ipod_mini2g_d5fc0e4cb3_rockbox.zip

These played a track from the File Browser without freezing (I only let them play for about 20 seconds):
Ipod_mini2g_95f4accf45_rockbox.zip
ipod_mini2g_801260dd79_HEAD_rockbox.zip
ipod_mini2g_0847bcc110_.zip

[iPodVT]

That set of four tests were all clean-installed.

Now I have used Rockbox Utility to update my full installation of 95f4accf45 with all of my configs etc to 801260dd79 and tracks are playing without freezing.

When you have a free moment can you please give us a word or two of explanation regarding what the problem was?

Thanks.

[Bilgus]

What this was was a subtle bug with our lang_id system
it has macros P2STR P2ID and ID2P
P2STR takes any old string and passes it thru but if its a lang string it grabs it from the lang table

P2ID and ID2P do similar conversions

well the way it is used it references an 'invalid' memory location to denote lang strings
on a bunch of the devices this was address '0'

that patch you referenced used P2STR(text) to voice user supplied entries
I caught the end result which was a NULL string referencing LANG_ID 0 and moved the 'Rockbox' string there
Ya know free advertising?

But Speachy correctly identified the flip side of it which is the crash you experienced

so to fix it he put the devices using address 0 at address 0 + sizeof(*char) preserving the sanctity of NULL and adding 100-400 bytes to all the bins

[iPodVT]

Thanks for the explanation.

I'm going to name my next album 'The Sanctity of NULL'.