Swagger API makes server to fail on panic

[VitSimon]

Description

When I use swagger page call:
http://localhost:3000/api/swagger#/issue/issueCreateIssue

operation is successful but server is terminated on panic.

JSONInput.txt
CLIOutput.txt

Logs

2022/11/19 11:47:46 ...s/graceful/server.go:62:NewServer() [I] [6378b452-46] Starting new Web server: tcp:0.0.0.0:3000 on PID: 1300
2022/11/19 11:47:58 router: completed POST /api/v1/repos/admin1/GRPCBlueprint/issues for 127.0.0.1:64688, 201 Created in 65.6ms @ repo/issue.go:562(repo.CreateIssue)
panic: unaligned 64-bit atomic operation

goroutine 1098 [running]:
runtime/internal/atomic.panicUnaligned()
        /usr/local/go/src/runtime/internal/atomic/unaligned.go:8 +0x2d
runtime/internal/atomic.Xadd64(0x1d210df4, 0x8)
        /usr/local/go/src/runtime/internal/atomic/atomic_386.s:125 +0x11
github.com/blevesearch/zapx/v15.(*chunkedContentCoder).incrementBytesWritten(...)
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/contentcoder.go:114
github.com/blevesearch/zapx/v15.(*chunkedContentCoder).flushContents(0x1d210d80)
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/contentcoder.go:143 +0x24a
github.com/blevesearch/zapx/v15.(*chunkedContentCoder).Close(...)
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/contentcoder.go:110
github.com/blevesearch/zapx/v15.(*interim).writeDicts(0x1f7c0f00)
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/new.go:803 +0x1443
github.com/blevesearch/zapx/v15.(*interim).convert(0x1f7c0f00)
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/new.go:277 +0x2b5
github.com/blevesearch/zapx/v15.(*ZapPlugin).newWithChunkMode(0x5b887a4, {0x1e00d7a8, 0x1, 0x1}, 0x402)
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/new.go:72 +0x18b
github.com/blevesearch/zapx/v15.(*ZapPlugin).New(0x5b887a4, {0x1e00d7a8, 0x1, 0x1})
        /go/pkg/mod/github.com/blevesearch/zapx/[email protected]/new.go:47 +0x43
github.com/blevesearch/bleve/v2/index/scorch.(*Scorch).Batch(0x1c937000, 0x1f7dea70)
        /go/pkg/mod/github.com/blevesearch/bleve/[email protected]/index/scorch/scorch.go:411 +0x5ff
github.com/blevesearch/bleve/v2.(*indexImpl).Batch(0x1bb045a0, 0x1e715d40)
        /go/pkg/mod/github.com/blevesearch/bleve/[email protected]/index_impl.go:309 +0xa0
code.gitea.io/gitea/modules/indexer/bleve.(*FlushingBatch).Flush(0x1adf1dc4)
        /source/modules/indexer/bleve/batch.go:53 +0x32
code.gitea.io/gitea/modules/indexer/issues.(*BleveIndexer).Index(0x1be45720, {0x1e00d6f0, 0x1, 0x1})
        /source/modules/indexer/issues/bleve.go:226 +0x183
code.gitea.io/gitea/modules/indexer/issues.InitIssueIndexer.func1({0x1b942640, 0x1, 0x14})
        /source/modules/indexer/issues/indexer.go:147 +0x5b1
code.gitea.io/gitea/modules/queue.NewByteFIFOQueue.func1({0x1b942640, 0x1, 0x14})
        /source/modules/queue/queue_bytefifo.go:71 +0x3e
code.gitea.io/gitea/modules/queue.(*WorkerPool).doWork(0x1d4f85a0, {0x45bdc20, 0x1d573540})
        /source/modules/queue/workerpool.go:564 +0x69f
code.gitea.io/gitea/modules/queue.(*WorkerPool).addWorkers.func1()
        /source/modules/queue/workerpool.go:312 +0x5d
created by code.gitea.io/gitea/modules/queue.(*WorkerPool).addWorkers
        /source/modules/queue/workerpool.go:310 +0x25

Gitea Version

1.18.0+rc0-23-gaf8b2250c

Can you reproduce the bug on the Gitea demo site?

No

Log Gist

No response

Screenshots

I cannot verify it on web site.

Git Version

git version 2.33.0.windows.2

Operating System

Microsoft Windows [Version 10.0.19044.2251]

How are you running Gitea?

Gitea downloaded from here:
https://dl.gitea.io/gitea/1.18

run via cmd command in console.

Database

SQLite

[VitSimon]

I am starting it without admin rights elevation

[yardenshoham]

Does your executable's architecture match your own?

[wxiaoguang]

It could be a bug in blevesearch for 32bit code.

Golang i386 code can not do atomic operations on unaligned 64bit fields.

So, use 64bit (amd64) release and everything will be fine.

[zeripath]

Looking at the logs we see a panic due to an unaligned atomic.

So I think we need to know a bit more about your architecture. As @yardenshoham says are you running the correct version for your architecture?

If so, if we look at https://github.com/blevesearch/zapx/blame/master/contentcoder.go. We see this:

type chunkedContentCoder struct {
	final     []byte
	chunkSize uint64
	currChunk uint64
	chunkLens []uint64

	w                io.Writer
	progressiveWrite bool

	chunkMetaBuf bytes.Buffer
	chunkBuf     bytes.Buffer

	chunkMeta []MetaData

	compressed []byte // temp buf for snappy compression

	// atomic access to this variable
	bytesWritten uint64
}

Now the bytesWritten here is used as an atomic, but if we look at the rest of the fields in the struct there's a bool and 2 bytes.Buffer above it. Which I could imagine might cause it to be misaligned.

If you're definitely using the correct architecture then this would constitute a bug in https://github.com/blevesearch/zapx/

[zeripath]

See: blevesearch/zapx#147

[VitSimon]

Hi, thanks.

I am using gitea-1.18-gogit-windows-4.0-386. I will try amd64 then. I think I have 64bit system. Then I will pass my result here.

[VitSimon]

@wxiaoguang , @zeripath , @yardenshoham :

I have downloaded amd64 package and tried the same action several times on previously existing database.
Behavior I have reported doesn't appears any more. Thanks again.

[VitSimon]

configuration problem on my side.