Date: Mon, 9 Jun 2014 14:28:32 -0400 From: Mark Johnston <markj@freebsd.org> To: "Sevan / Venture37" <venture37@gmail.com> Cc: "freebsd-dtrace@freebsd.org" <freebsd-dtrace@freebsd.org> Subject: Re: DTrace probes for python 2.7.7 Message-ID: <CAMw1wOzE9An3Yz9XJh4xezoGk_UMJAxS=nJdj5Z4UquvverOgw@mail.gmail.com> In-Reply-To: <CA%2BU3Mf7WJbGbqEAF7Zm77TYDAtSKbuU2pq=YwR=obtnfupLtYQ@mail.gmail.com> References: <CA%2BU3Mf7_t3GuRMS0YicRb03ti56LHg_DD=iC=%2BD_wEj2SPNwyg@mail.gmail.com> <CA%2BU3Mf4ojD08wsWvRDkRDBmnk=eoJZCOD21hAMx1BC14_Kp5ZQ@mail.gmail.com> <CA%2BU3Mf7WJbGbqEAF7Zm77TYDAtSKbuU2pq=YwR=obtnfupLtYQ@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jun 9, 2014 at 2:16 PM, Sevan / Venture37 <venture37@gmail.com> wro= te: > On 4 June 2014 20:15, Sevan / Venture37 <venture37@gmail.com> wrote: >> dtrace: failed to compile script Include/pydtrace.d: line 26: typedef >> redeclared: __uint8_t > > On i386 the build fails differently with r266655 snapshot from 20140525 > > =E2=80=9Cfailed to compile script Include/pydtrace.d: =E2=80=9C/usr/lib/d= trace/ip.d=E2=80=9D > line 2: type redeclared: struct devinfo" > > There is of course an inconsistent. > > The original error regarding the redeclaration of __uint8_t was from a > AMD64 host running r266972. > > Installing the AMD64 r266655 snapshot & retrying the build results in, > > failed to compile script Include/pydtrace.d: > =E2=80=9C/usr/lib/dtrace/unistd.d=E2=80=9D, line 135: syntax error near = =E2=80=9Ccsinfo_t=E2=80=9D > > Can anyone shed some light on what's going on. This is probably because the build is running dtrace -G without the dtrace kernel modules loaded. This shouldn't be a problem, but currently is because dtrace(1) parses the scripts in /usr/lib/dtrace and errors out if dtrace isn't available in the kernel. Things like "#pragma depends_on" don't work, and there seem to be bugs in the error reporting of libdtrace; one ends up with the error messages above. I'd like to fix this by changing dtrace -G to not process scripts in the libdir, but I'm not yet sure what the implications of this are. I will ask on the dtrace-discuss list soon. In the meantime, a somewhat unsatisfactory workaround would be to just kldload dtraceall before starting the build. -Mark
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAMw1wOzE9An3Yz9XJh4xezoGk_UMJAxS=nJdj5Z4UquvverOgw>