Date: Mon, 29 Aug 2016 10:07:15 +0300 From: Konstantin Belousov <kostikbel@gmail.com> To: John Baldwin <jhb@freebsd.org> Cc: Andrey Chernov <ache@freebsd.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r304928 - in head/lib/libc: amd64/sys i386/sys sys Message-ID: <20160829070715.GQ83214@kib.kiev.ua> In-Reply-To: <20160829065813.GP83214@kib.kiev.ua> References: <201608272303.u7RN3N0D078505@repo.freebsd.org> <80ad9e03-74bc-8c99-666f-787772bef2b9@freebsd.org> <20160828015210.GI83214@kib.kiev.ua> <1595604.93PBdSz0kX@ralph.baldwin.cx> <20160829065813.GP83214@kib.kiev.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Aug 29, 2016 at 09:58:13AM +0300, Konstantin Belousov wrote:
> On Sun, Aug 28, 2016 at 04:09:51PM -0700, John Baldwin wrote:
> > OTOH, given that we explicitly documented it as not being true, I suspect
> > any applications that are using ptrace() are going off the documentation, not
> > the implementation artifact. Note that Linux's ptrace() documents the same
> > requirement as before this change (caller is required to clear errno), so I
> > doubt there is any actual software out there that expects the
> > FreeBSD-specific behavior. Given that and the extra maintenance overhead of
> > having to dink with errno in assembly on X architectures, I'd rather we keep
> > the old language in the manpage and remove the 'errno' frobbing in the system
> > call wrappers. To be honest, my first response to this commit was one of
> > surprise that we modify errno directly as that is inconsistent with other
> > system calls. (I haven't looked to see if any other system call wrappers
> > modify errno for non-error cases.)
>
> The problematic calls are PT_PEEK_I and PT_PEEK_D, as far as I understand.
>
> I dug into the ptrace(2) consumers, I found a lot of things using
> it which I would not expect to use, besides usual suspects of gdb
> lldb libunwind reptyr etc. Most surprising was that even high-profile
> consumers including gdb sometimes fail to check errno after PT_PEEK. On
> the other hand, I did not found a case in gdb where errno is checked
> after PT_PEEK but not zeroed before the syscall.
>
> I almost agreed with you after the reading, but then I decided to look
> into glibc just in case. What I found there is really fascinating.
> From glibc/sysdeps/unix/sysv/linux:
> res = INLINE_SYSCALL (ptrace, 4, request, pid, addr, data);
> if (res >= 0 && request > 0 && request < 4)
> {
> __set_errno (0);
> return ret;
> }
> #define PTRACE_PEEKTEXT 1
> #define PTRACE_PEEKDATA 2
> #define PTRACE_PEEKUSR 3
>
> In the end, I might consider changing the ptrace wrappers into
> consolidated C source, it would look like that
>
> int
> ptrace(int request, pid_t pid, caddr_t addr, int data)
> {
>
> errno = 0;
> return (__sys_ptrace(request, pid, addr, data));
> }
And Solaris libc, where ptrace() is the wrapper around procfs, starts its
implementation this way:
usr/src/lib/libc/i386/sys/ptrace.c
/*
* Process the request.
*/
errno = 0;
switch (request) {
case 1: /* PTRACE_PEEKTEXT */
case 2: /* PTRACE_PEEKDATA */
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20160829070715.GQ83214>