-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256
On 2019-01-12 7:23 p.m., Michael Biebl wrote:
> Hi Jussi > > On Sun, 2 Dec 2018 00:23:09 +0200 Jussi Pakkanen
wrote: > >> The correct way to enable PIE
executables with Meson is to remove all >> your own compiler and link
flags related to pie and
Hi Jussi
On Sun, 2 Dec 2018 00:23:09 +0200 Jussi Pakkanen wrote:
> The correct way to enable PIE executables with Meson is to remove all
> your own compiler and link flags related to pie and set b_pie to true.
> This is a fairly new feature and probably the reason why systemd is
> not using it.
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256
Hi Michael,
On 2019-01-10 7:09 a.m., Michael Biebl wrote:
> Dave, can you test attached patch on hppa and report back with the results?
> Version 240-3 built successfully with patch on hppa.
Dave
- --
John David Anglin dave.ang...@bell.net
-
On 2019-01-10 10:46 a.m., Michael Biebl wrote:
>>> Dave, can you test attached patch on hppa and report back with the results?
>> Your fix looks correct to me. I'll try to test in the next day or two
>> on hppa. Does it work on x86?
> I've built the package on amd64 with this patch applied.
> I q
Am 10.01.19 um 16:13 schrieb John David Anglin:
> Hi,
>
> The merging of PIC/PIE options in gcc-8 was fixed by this change:
> https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=267610
>
> A new build of gcc-8 with the above change was upload yesterday and
> systemd 240-3 built
> successfully
Hi,
The merging of PIC/PIE options in gcc-8 was fixed by this change:
https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=267610
A new build of gcc-8 with the above change was upload yesterday and
systemd 240-3 built
successfully on hppa without having to disable LTO:
https://buildd.debian.org
Hi Dave, hi Jussi
On Sat, 1 Dec 2018 17:15:41 -0500 John David Anglin
wrote:
> On 2018-12-01 4:50 p.m., Jussi Pakkanen wrote:
> > Thus it would seem that this is not a bug in Meson, but instead in
> > systemd's build setup as the pie arguments are added by the latter.
> I agree but Michael doesn'
On Sun, Dec 2, 2018 at 6:34 PM John David Anglin wrote:
> However, we didn't get PIE executables. The current version of meson is
> 0.48.2-1 and b_pie was introduced in 0.49.0.
0.49.0 will be released next Sunday so the issue will get fixed in about a week.
On 2018-12-01 5:23 p.m., Jussi Pakkanen wrote:
The correct way to enable PIE executables with Meson is to remove all
your own compiler and link flags related to pie and set b_pie to true.
This is a fairly new feature and probably the reason why systemd is
not using it.
Removing the compiler and l
On 2018-12-01 4:50 p.m., Jussi Pakkanen wrote:
Thus it would seem that this is not a bug in Meson, but instead in
systemd's build setup as the pie arguments are added by the latter.
I agree but Michael doesn't have a clear idea how to fix the issue.
Would the "b_pie" option help?
Regards,
Dave
On Sun, Dec 2, 2018 at 12:15 AM John David Anglin wrote:
> > Thus it would seem that this is not a bug in Meson, but instead in
> > systemd's build setup as the pie arguments are added by the latter.
> I agree but Michael doesn't have a clear idea how to fix the issue.
> Would the "b_pie" option
On Sat, Nov 24, 2018 at 6:09 PM John David Anglin wrote:
> My understanding is -fPIE should be used when creating objects for a
> PIE executable. -fPIC should be used when creating objects for a shared
> library. Only one is recorded when -flto is specified.
I reviewed the code and we only add
Source: meson
Severity: normal
Dear Maintainer,
The systemd package fails to build on hppa and x32. This is caused by
both -fPIE and -fPIC options being added to various compilation commands.
Only -fPIC is being recorded in the LTO options section of the object.
The gcc-8 LTO plugin merges -fPIC
13 matches
Mail list logo
