Yes, I think you are right about the case in my sample code. The code doesn't show exactly what I'm concerned about. I just wanna know whether a write to a global pointer in 64-bit machine an atomic operation or not.
在 2017年1月9日星期一 UTC+8上午6:22:12，Caleb Doxsey写道： > > Shouldn't this particular case be ok? From the memory doc: > https://golang.org/ref/mem > > The go statement that starts a new goroutine happens before the >> goroutine's execution begins. > > > So the write has to happen before the goroutine starts. At least that's > what the example indicates: > > >> For example, in this program: > > > > var a string > > func f() { > > print(a) > > } > > func hello() { > > a = "hello, world" > > go f() > > } > > > >> calling hello will print "hello, world" at some point in the future >> (perhaps after hello has returned). > > > On Sunday, January 8, 2017 at 1:17:56 PM UTC-5, 陈诚 wrote: > >> Is the size of a pointer value 32 bits or 64 bits in golang when build >> with `GOARCH=amd64` option specified and running on 64-bit OS? >> If it's 64-bit size, is a global pointer value 8-byte aligned in memory >> so that a read or write operation of that pointer value is carried out >> atomically? >> For example, in the following code, is it possible that the global >> pointer p is only partially updated when the read goroutine read the >> pointer? >> var p *int >> >> void main() { >> i := 1 >> p = &i >> go func() { fmt.Println(*p) } () >> } >> >> The scenario I'm concerning is that there is only one write but multiple >> reads on a global pointer value, and reading of an old value of the pointer >> is not important. Thanks in advance! >> > -- 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. For more options, visit https://groups.google.com/d/optout.
