On Fri, 14 Jan 2022 12:52:11 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
>> Very trivial fix to a handle/memory leak. >> >> JDK-8215624 added parallel heap iteration to both `jmap -histo` and `jcmd >> GC.class_histogram`. When called with an explicit file and an invalid >> argument for number of threads, it leaks the file (bit of memory and a >> handle). >> >> Reproduce with: >> >> `jmap -histo:parallel=notanumber,file=xx.txt` >> >> Can only be reproduced with jmap. jcmd is safe, arguments are handled >> correctly in shared code. > > This is not as trivial, AFAICS. Note that the existing code `delete fs` after > checking `if (os != NULL && os != out)`. Does it mean this patch can > effectively `delete out`? Or can it call `delete nullptr`? Thanks, @shipilev and @sspitsyn ! ------------- PR: https://git.openjdk.java.net/jdk/pull/7078
