Crash in socket.go attempting to update BLEFS #11

New Issue

brennana · 2022-03-06T01:31:17Z

brennana commented

2022-03-06 01:31:17 +00:00

Sometimes, when attempting reconnection, itd crashes during the BLEFS update phase of handleConnection. In my particular case, it seems to be because the watch lost connection again before the FS update attempt (apparently not far away enough to induce the pair-code issue, however).

In any case, if the dev.FS() operation fails, there is a chance that a segfault will arise:

Mar 05 17:38:33 pinephonepro itd[47055]: 5:38PM WRN Error updating BLE filesystem error="not connected"
Mar 05 17:38:33 pinephonepro itd[47055]: panic: runtime error: invalid memory address or nil pointer dereference
Mar 05 17:38:33 pinephonepro itd[47055]: [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x3705ac]
Mar 05 17:38:33 pinephonepro itd[47055]: goroutine 57 [running]:
Mar 05 17:38:33 pinephonepro itd[47055]: main.handleConnection({0x4d75d8, 0x400010c018}, 0x400018cb00, 0x400015c408)
Mar 05 17:38:33 pinephonepro itd[47055]:         /home/andy/.cache/yay/itd-git/src/itd/socket.go:119 +0x10c
Mar 05 17:38:33 pinephonepro itd[47055]: created by main.startSocket.func1
Mar 05 17:38:33 pinephonepro itd[47055]:         /home/andy/.cache/yay/itd-git/src/itd/socket.go:97 +0x50

Sometimes, when attempting reconnection, itd crashes during the BLEFS update phase of `handleConnection`. In my particular case, it seems to be because the watch lost connection again before the FS update attempt (apparently not far away enough to induce the pair-code issue, however). In any case, if the `dev.FS()` operation fails, there is a chance that a segfault will arise: ``` Mar 05 17:38:33 pinephonepro itd[47055]: 5:38PM WRN Error updating BLE filesystem error="not connected" Mar 05 17:38:33 pinephonepro itd[47055]: panic: runtime error: invalid memory address or nil pointer dereference Mar 05 17:38:33 pinephonepro itd[47055]: [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x3705ac] Mar 05 17:38:33 pinephonepro itd[47055]: goroutine 57 [running]: Mar 05 17:38:33 pinephonepro itd[47055]: main.handleConnection({0x4d75d8, 0x400010c018}, 0x400018cb00, 0x400015c408) Mar 05 17:38:33 pinephonepro itd[47055]: /home/andy/.cache/yay/itd-git/src/itd/socket.go:119 +0x10c Mar 05 17:38:33 pinephonepro itd[47055]: created by main.startSocket.func1 Mar 05 17:38:33 pinephonepro itd[47055]: /home/andy/.cache/yay/itd-git/src/itd/socket.go:97 +0x50 ```

Elara6331 referenced this issue from a commit

2022-03-08 16:32:40 +00:00

Make sure fs is only updated if dev.FS() succeeds (#11)

Elara6331 commented

2022-03-08 16:33:21 +00:00

It should be fixed now. Please test and confirm. Thank you for finding these bugs.

Elara6331 added the

Bug

Waiting For Info

labels 2022-03-08 16:33:30 +00:00

brennana commented

2022-03-08 23:37:03 +00:00

Rebuilt and tested with the latest commit, and upon receiving the warning from BLEFS update, the daemon now continues gracefully instead of segfaulting.

Thanks for building such a robust Pinetime daemon (and fixing these issues so quickly)! :D

Rebuilt and tested with the latest commit, and upon receiving the warning from BLEFS update, the daemon now continues gracefully instead of segfaulting. Thanks for building such a robust Pinetime daemon (and fixing these issues so quickly)! :D

brennana closed this issue

2022-03-08 23:37:03 +00:00

Elara6331 removed the

Waiting For Info

label 2022-03-09 04:02:01 +00:00

Elara6331 referenced this issue from a commit

2023-04-21 02:08:29 +00:00

Make sure fs is only updated if dev.FS() succeeds (#11)

Sign in to join this conversation.