Date: Mon, 31 Oct 2005 07:44:55 +0800 From: David Xu <davidxu@freebsd.org> To: Alfred Perlstein <alfred@freebsd.org> Cc: alc@freebsd.org, cvs-src@FreeBSD.org, src-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/sys aio.h signal.h signalvar.h src/sys/kern syscalls.master vfs_aio.c Message-ID: <43655AF7.8070707@freebsd.org> In-Reply-To: <20051030134702.GK32289@elvis.mu.org> References: <200510300212.j9U2CnWl089007@repoman.freebsd.org> <20051030134702.GK32289@elvis.mu.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Existing binary code needn't be recompiled, it should work as before. if an AIO syscall needs sigevent structure, it already has a new syscall number, and old syscalls will think passed sigevent is an osigevent. I am open for adding pad fields to aiocb and sigevent, how many padding do you want to add, give some diffs for us ? :-) David Xu Alfred Perlstein wrote: >Does this modify the size/layout of the struct sigevent and aiocb >to the point where apps need a recompile? If so we >should we have compat functions for AIO? Ei, struct "osigevent" >and "oaiocb" and corresponding compat functions for AIO? > >I understand that it's annoying but AIO has been an interface around >since 4.x. > >Now might also be a good time to add padding to aiocb and possibly >sigevent. > >-Alfred > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?43655AF7.8070707>