SEGV during (*Tx).rollback

[tmm1]

unexpected fault address 0x2ab91044
fatal error: fault
[signal SIGSEGV: segmentation violation code=0x1 addr=0x2ab91044 pc=0x3209d8]

goroutine 7365 [running]:
runtime.throw(0xd9fe37, 0x5)
	runtime/panic.go:774 +0x5c fp=0x59d3a38 sp=0x59d3a24 pc=0x42374
runtime.sigpanic()
	runtime/signal_unix.go:401 +0x310 fp=0x59d3a50 sp=0x59d3a38 pc=0x5928c
github.com/etcd-io/bbolt.(*DB).meta(0x3d8cc80, 0x721)
	github.com/etcd-io/bbolt@v1.3.3/db.go:901 +0x1c fp=0x59d3a6c sp=0x59d3a54 pc=0x3209d8
github.com/etcd-io/bbolt.(*DB).hasSyncedFreelist(...)
	github.com/etcd-io/bbolt@v1.3.3/db.go:323
github.com/etcd-io/bbolt.(*Tx).rollback(0x51e4b00)
	github.com/etcd-io/bbolt@v1.3.3/tx.go:279 +0x68 fp=0x59d3a84 sp=0x59d3a6c pc=0x329ea8
github.com/etcd-io/bbolt.(*Tx).Commit(0x51e4b00, 0x3c63b88, 0x2)
	github.com/etcd-io/bbolt@v1.3.3/tx.go:161 +0x430 fp=0x59d3b24 sp=0x59d3a84 pc=0x3298d8
github.com/blevesearch/bleve/index/scorch.(*Scorch).persistSnapshotDirect(0x3ca1c00, 0x3cc36e0, 0x0, 0x0)
	github.com/blevesearch/bleve@v0.8.2-0.20191010234049-157461a2aeb6/index/scorch/persister.go:580 +0xfa0 fp=0x59d3de4 sp=0x59d3b24 pc=0x705e1c
github.com/blevesearch/bleve/index/scorch.(*Scorch).persistSnapshot(0x3ca1c00, 0x3cc36e0, 0x416d790, 0x3ed8, 0x1ac5760)
	github.com/blevesearch/bleve@v0.8.2-0.20191010234049-157461a2aeb6/index/scorch/persister.go:349 +0x74 fp=0x59d3dfc sp=0x59d3de4 pc=0x7045f8
github.com/blevesearch/bleve/index/scorch.(*Scorch).persisterLoop(0x3ca1c00)
	github.com/blevesearch/bleve@v0.8.2-0.20191010234049-157461a2aeb6/index/scorch/persister.go:141 +0x6f8 fp=0x59d3fe4 sp=0x59d3dfc pc=0x703440
runtime.goexit()
	runtime/asm_arm.s:868 +0x4 fp=0x59d3fe4 sp=0x59d3fe4 pc=0x747a8
created by github.com/blevesearch/bleve/index/scorch.(*Scorch).Open
	github.com/blevesearch/bleve@v0.8.2-0.20191010234049-157461a2aeb6/index/scorch/scorch.go:170 +0xc8

I believe this happened during low memory pressure on the device.

[leakybits]

We just hit the same error. Seems there was already an issue opened here (which was closed because there was no code that could reproduce it) and here. Are you able to reliably reproduce this?

[ahrtr]

Fixed in #362.

Please anyone let me know if you still see the panic.