> This looks right and reduces the stat usages in libjava down to one > remaining case (ProcessHandlerImpl_linux.c).
Hi Alan, Do you want me to do this one too for completeness? Although as it's calling stat() on the /proc/<pid> directory I don't think it's possible to hit the EOVERFLOW case. Thanks, Nick > -----Original Message----- > From: Alan Bateman <alan.bate...@oracle.com> > Sent: 23 November 2018 19:01 > To: Nick Gasson <nick.gas...@arm.com>; Magnus Ihse Bursie > <magnus.ihse.bur...@oracle.com>; build-dev <build-...@openjdk.java.net>; > core-libs-dev@openjdk.java.net > Cc: nd <n...@arm.com> > Subject: Re: RFR: 8214077: test java/io/File/SetLastModified.java fails on > ARM32 > > On 23/11/2018 09:37, Nick Gasson wrote: > > Hi Alan, > > > > I've done this here: > > > > http://cr.openjdk.java.net/~njian/8214077/webrev.2/ > This looks right and reduces the stat usages in libjava down to one > remaining case (ProcessHandlerImpl_linux.c). > > > > > I'm a bit unsure about #ifndef MACOSX - some existing files guard > > the stat64 #define with this (e.g. UnixFileSystem_md.c) and some > > don't (e.g. libnio/ch/FileDispatcherImpl.c). I don't have access > > to an OS X machine to test, but I guess it doesn't matter too > > much as AFAIK Apple haven't supported 32-bit for a long time? > > > Some sources are using _ALLBSD_SOURCE, I think this dates back to the > Apple/BSD patch (went into 7u4 and 8). So there is a bit of consistency > that should be cleaned up some time. Apple may have had a 32-bit build > of the JDK many years ago but the changes that were into OpenJDK were > for building on 64-bit only. > > -Alan