Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-22 Thread Richard Z
On Fri, Feb 20, 2015 at 07:28:50PM +, Peter Robinson wrote: On Fri, Feb 20, 2015 at 6:55 PM, Till Maas opensou...@till.name wrote: On Fri, Feb 20, 2015 at 05:21:59PM +, Peter Robinson wrote: I've never argumented against the goal that web browser or all network aware services

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-21 Thread Till Maas
On Fri, Feb 20, 2015 at 07:28:50PM +, Peter Robinson wrote: On Fri, Feb 20, 2015 at 6:55 PM, Till Maas opensou...@till.name wrote: On Fri, Feb 20, 2015 at 05:21:59PM +, Peter Robinson wrote: How is a PDF with a binary payload any different? Sounds like we need to be running pdf

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Peter Robinson
I've never argumented against the goal that web browser or all network aware services should be PIEs, after all, why would we (Ulrich Drepper and myself) add the PIE support into the toolchain otherwise? I'm just not convinced most of the unpriviledged programs should be PIEs. Thanks to e.g.

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Reindl Harald
Am 20.02.2015 um 18:21 schrieb Peter Robinson: I've never argumented against the goal that web browser or all network aware services should be PIEs, after all, why would we (Ulrich Drepper and myself) add the PIE support into the toolchain otherwise? I'm just not convinced most of the

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Ralf Corsepius
On 02/20/2015 07:55 PM, Till Maas wrote: than the technical change to implement it, there's no mention that it will have an impact on performance, with numbers to back it up, across the three primary architectures. So how much performance impact is acceptable? None - Seriously, people will

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Josh Boyer
On Fri, Feb 20, 2015 at 12:21 PM, Peter Robinson pbrobin...@gmail.com wrote: Also I've seen no performance analysis across all three architectures to see the impact. I'll happily send you an XO-1 to test on (our lowest supported device on i686 and also one of our most widely deployed Fedora

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Till Maas
On Fri, Feb 20, 2015 at 05:21:59PM +, Peter Robinson wrote: I've never argumented against the goal that web browser or all network aware services should be PIEs, after all, why would we (Ulrich Drepper and myself) add the PIE support into the toolchain otherwise? I'm just not

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Till Maas
On Fri, Feb 20, 2015 at 01:40:30PM -0500, Josh Boyer wrote: (XO-4). There's the whole Android angle now to, which frankly makes a lot of sense for them. Even the latest OLPC OS development release is based on Fedora 20 and is targeted at the XO-4 hardware. Android already uses ASLR, see eg

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Peter Robinson
On Fri, Feb 20, 2015 at 6:40 PM, Josh Boyer jwbo...@fedoraproject.org wrote: On Fri, Feb 20, 2015 at 12:21 PM, Peter Robinson pbrobin...@gmail.com wrote: Also I've seen no performance analysis across all three architectures to see the impact. I'll happily send you an XO-1 to test on (our

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Peter Robinson
On Fri, Feb 20, 2015 at 6:55 PM, Till Maas opensou...@till.name wrote: On Fri, Feb 20, 2015 at 05:21:59PM +, Peter Robinson wrote: I've never argumented against the goal that web browser or all network aware services should be PIEs, after all, why would we (Ulrich Drepper and

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-20 Thread Reindl Harald
Am 20.02.2015 um 20:28 schrieb Peter Robinson: On Fri, Feb 20, 2015 at 6:55 PM, Till Maas opensou...@till.name wrote: So how much performance impact is acceptable? Well you've not documented any of the impact so how can we discuss that? We have no idea if the impact is going to be 0.1% 1% or

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 09:14:32AM +, Richard W.M. Jones wrote: https://bugzilla.redhat.com/show_bug.cgi?id=1194167 Basically everything in Rawhide which uses the normal RPM opt flags will now have to be compiled with -fPIC, otherwise you get errors like: /usr/bin/ld:

So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
https://bugzilla.redhat.com/show_bug.cgi?id=1194167 Basically everything in Rawhide which uses the normal RPM opt flags will now have to be compiled with -fPIC, otherwise you get errors like: /usr/bin/ld: /tmp/ccqyK5ia.o: relocation R_X86_64_32S against `virConnectOpen' can not be used when

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 07:58:10PM +0100, Reindl Harald wrote: Am 19.02.2015 um 19:48 schrieb Till Maas: On Thu, Feb 19, 2015 at 07:07:45PM +0100, Jakub Jelinek wrote: Even on x86_64 it was quite a measurable slowdown last time I've benchmarked it, now in F22+ we might have smaller

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Reindl Harald
Am 19.02.2015 um 20:15 schrieb Jakub Jelinek: On Thu, Feb 19, 2015 at 07:58:10PM +0100, Reindl Harald wrote: Am 19.02.2015 um 19:48 schrieb Till Maas: On Thu, Feb 19, 2015 at 07:07:45PM +0100, Jakub Jelinek wrote: Even on x86_64 it was quite a measurable slowdown last time I've benchmarked

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Dave Johansen
On Thu, Feb 19, 2015 at 12:34 PM, Till Maas opensou...@till.name wrote: On Thu, Feb 19, 2015 at 08:15:19PM +0100, Jakub Jelinek wrote: I've never argumented against the goal that web browser or all network aware services should be PIEs, after all, why would we (Ulrich Drepper and myself)

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Till Maas
On Thu, Feb 19, 2015 at 08:15:19PM +0100, Jakub Jelinek wrote: I've never argumented against the goal that web browser or all network aware services should be PIEs, after all, why would we (Ulrich Drepper and myself) add the PIE support into the toolchain otherwise? I'm just not convinced

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
On Thu, Feb 19, 2015 at 11:35:17AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 10:30:50AM +, Richard W.M. Jones wrote: info gcc, of course yes. -DPIC is not documented at all, and the various pie/pic options are obscure to say the least. Why should -DPIC be documented? -D is

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 10:37:46AM +, Richard W.M. Jones wrote: On Thu, Feb 19, 2015 at 11:35:17AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 10:30:50AM +, Richard W.M. Jones wrote: info gcc, of course yes. -DPIC is not documented at all, and the various pie/pic options

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
On Thu, Feb 19, 2015 at 10:22:03AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 09:14:32AM +, Richard W.M. Jones wrote: https://bugzilla.redhat.com/show_bug.cgi?id=1194167 Basically everything in Rawhide which uses the normal RPM opt flags will now have to be compiled with

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
I'm still no closer to being able to fix the problem. I have to add -fPIE (or is that -pie or -fpie or -DPIE) to every executable? Upstream? Will that break on some platforms/architectures? And indeed what is the difference between -fPIE / -pie / -fpie / -fPIC / -fpic / -DPIC / etc? Where is

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
On Thu, Feb 19, 2015 at 11:23:18AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 10:11:07AM +, Richard W.M. Jones wrote: I'm still no closer to being able to fix the problem. I have to add -fPIE (or is that -pie or -fpie or -DPIE) to every executable? Upstream? Will that

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
And in another case I had (qemu), it's not predictable. If you just compile part of qemu twice, the first time it gives the error and the second time not. I had to add this hack to qemu.spec: http://pkgs.fedoraproject.org/cgit/qemu.git/commit/?id=6c3741c2769a21542a34716fa9188e520887a803 Rich.

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
On Thu, Feb 19, 2015 at 10:42:02AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 09:37:16AM +, Richard W.M. Jones wrote: The thing is, I'm not adding -specs=/usr/lib/rpm/redhat/redhat-hardened-ld explicitly in the real program. It's being added to everything by something in RPM.

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 09:51:33AM +, Richard W.M. Jones wrote: There is definitely new/different behaviour in Rawhide, and recently. Also I was only able to see the new behaviour by updating from gcc 4.x - gcc 5. ie. Updating redhat-rpm-config isn't what causes the problem. Well, gcc

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 10:11:07AM +, Richard W.M. Jones wrote: I'm still no closer to being able to fix the problem. I have to add -fPIE (or is that -pie or -fpie or -DPIE) to every executable? Upstream? Will that break on some platforms/architectures? It really should be just about

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 10:30:50AM +, Richard W.M. Jones wrote: info gcc, of course yes. -DPIC is not documented at all, and the various pie/pic options are obscure to say the least. Why should -DPIC be documented? -D is documented. -DPIC means define macro PIC to 1. There is no magic

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
On Thu, Feb 19, 2015 at 11:45:01AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 10:37:46AM +, Richard W.M. Jones wrote: On Thu, Feb 19, 2015 at 11:35:17AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 10:30:50AM +, Richard W.M. Jones wrote: info gcc, of course yes.

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 09:37:16AM +, Richard W.M. Jones wrote: The thing is, I'm not adding -specs=/usr/lib/rpm/redhat/redhat-hardened-ld explicitly in the real program. It's being added to everything by something in RPM. I'm not exactly sure what, maybe %{configure}? So I don't know

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Richard W.M. Jones
On Thu, Feb 19, 2015 at 10:37:46AM +, Richard W.M. Jones wrote: On Thu, Feb 19, 2015 at 11:35:17AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 10:30:50AM +, Richard W.M. Jones wrote: info gcc, of course yes. -DPIC is not documented at all, and the various pie/pic options

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Till Maas
On Thu, Feb 19, 2015 at 07:07:45PM +0100, Jakub Jelinek wrote: Even on x86_64 it was quite a measurable slowdown last time I've benchmarked it, now in F22+ we might have smaller slowdown with the x86_64 copyreloc for Which packages are there that do not process untrusted data and are slowed

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Reindl Harald
Am 19.02.2015 um 19:48 schrieb Till Maas: On Thu, Feb 19, 2015 at 07:07:45PM +0100, Jakub Jelinek wrote: Even on x86_64 it was quite a measurable slowdown last time I've benchmarked it, now in F22+ we might have smaller slowdown with the x86_64 copyreloc for Which packages are there that do

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 07:48:30PM +0100, Till Maas wrote: On Thu, Feb 19, 2015 at 07:07:45PM +0100, Jakub Jelinek wrote: Even on x86_64 it was quite a measurable slowdown last time I've benchmarked it, now in F22+ we might have smaller slowdown with the x86_64 copyreloc for Which

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Christopher Meng
I'd like to know the performance loss estimated/examined by FESCO, or I think it's a bad idea on i686 systems, it's already slow enough. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct:

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Till Maas
On Thu, Feb 19, 2015 at 10:42:02AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 09:37:16AM +, Richard W.M. Jones wrote: The thing is, I'm not adding -specs=/usr/lib/rpm/redhat/redhat-hardened-ld explicitly in the real program. It's being added to everything by something in RPM.

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 06:00:41PM +0100, Till Maas wrote: On Thu, Feb 19, 2015 at 10:42:02AM +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 09:37:16AM +, Richard W.M. Jones wrote: The thing is, I'm not adding -specs=/usr/lib/rpm/redhat/redhat-hardened-ld explicitly in the real

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Adam Jackson
On Thu, 2015-02-19 at 18:05 +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 06:00:41PM +0100, Till Maas wrote: I plan to change this to 1 now in Rawhide for the upcoming Fedora 23 feature: https://bugzilla.redhat.com/show_bug.cgi?id=1192183 Since I did not get any feedback from the

Re: So everything in Rawhide must be compiled with -fPIC?

2015-02-19 Thread Jakub Jelinek
On Thu, Feb 19, 2015 at 01:02:29PM -0500, Adam Jackson wrote: On Thu, 2015-02-19 at 18:05 +0100, Jakub Jelinek wrote: On Thu, Feb 19, 2015 at 06:00:41PM +0100, Till Maas wrote: I plan to change this to 1 now in Rawhide for the upcoming Fedora 23 feature: