hi Axel, is not modifying `u.memoize.str` thread-safe? the len and the data point should become visible at same time?
在2024年3月16日星期六 UTC+8 06:29:06<Axel Wagner> 写道： > Have you tried running the code with the race detector enabled? I suspect > that you are concurrently modifying `u.memoize.str` by calling `u.String()` > from multiple goroutines. And the non-zero length of the string header > written by one goroutine becomes visible to the other one, before the > modification to the data pointer. > > On Fri, Mar 15, 2024 at 11:15 PM Ethan Reesor <ethan....@gmail.com> wrote: > >> From this CI job >> <https://gitlab.com/accumulatenetwork/accumulate/-/jobs/6398114923>: >> >> panic: runtime error: invalid memory address or nil pointer dereference >> [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x51d8b7] >> goroutine 1589381 [running]: >> strings.EqualFold({0xc000beec20?, 0x0?}, {0x0?, 0xacace7?}) >> /usr/local/go/src/strings/strings.go:1111 +0x37 >> gitlab.com/accumulatenetwork/accumulate/pkg/url.(*URL).Equal(0xc000a74e40?, >> 0xc00094c540) >> /builds/accumulatenetwork/accumulate/pkg/url/url.go:472 +0x10c >> >> This is in a docker container based on the go:1.22 image, so the panic >> appears to be happening here: >> >> func EqualFold(s, t string) bool { >> // ASCII fast path >> i := 0 >> for ; i < len(s) && i < len(t); i++ { >> sr := s[i] >> tr := t[i] // <-- line 1111 >> >> (*URL).Equal >> <https://gitlab.com/accumulatenetwork/accumulate/-/blob/5b1cb612d76d4163a101303e51a6fd352224cdab/pkg/url/url.go#L465> >> : >> >> func (u *URL) Equal(v *URL) bool { >> if u == v { >> return true >> } >> if u == nil || v == nil { >> return false >> } >> return strings.EqualFold(u.String(), v.String()) >> } >> >> (*URL).String >> <https://gitlab.com/accumulatenetwork/accumulate/-/blob/5b1cb612d76d4163a101303e51a6fd352224cdab/pkg/url/url.go#L240> >> : >> >> func (u *URL) String() string { >> if u.memoize.str != "" { >> return u.memoize.str >> } >> >> u.memoize.str = u.format(nil, true) >> return u.memoize.str >> } >> >> (*URL).format >> <https://gitlab.com/accumulatenetwork/accumulate/-/blob/5b1cb612d76d4163a101303e51a6fd352224cdab/pkg/url/url.go#L189> >> : >> >> func (u *URL) format(txid []byte, encode bool) string { >> var buf strings.Builder >> // ... write to the builder >> return buf.String() >> } >> >> How is this possible? Based on `addr=0x0` in the panic I think this is a >> nil pointer panic, as opposed to some other kind of segfault. The only way >> I can reproduce panic-on-string-index is with >> `(*reflect.StringHeader)(unsafe.Pointer(&s)).Data >> = 0`, but I don't see how that can be happening here. I'm saving the string >> but I'm not doing anything weird with it. And the string header is a value >> type so code that manipulates the returned string shouldn't modify the >> original. And I'm definitely not doing any kind of unsafe string >> manipulation like that in my code, anywhere. The only reference to unsafe >> anywhere in my code is for parameters for calling GetDiskFreeSpaceExW >> (Windows kernel32.dll call). >> >> -- >> You received this message because you are subscribed to the Google Groups >> "golang-nuts" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to golang-nuts...@googlegroups.com. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/golang-nuts/d6f6bb75-45e9-4a38-9bbd-d332e7f3e57cn%40googlegroups.com >> >> <https://groups.google.com/d/msgid/golang-nuts/d6f6bb75-45e9-4a38-9bbd-d332e7f3e57cn%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/31f77ff2-cf11-4b3e-9b14-874b6cc41da3n%40googlegroups.com.
