jrtc27 (James Clarke)
User

Projects

User does not belong to any projects.

User Details

User Since
Nov 1 2016, 4:54 AM (98 w, 6 d)

Recent Activity

Jul 15 2018

jrtc27 accepted D4959: driver: don't enable -Bsymbolic on unregisterised targets (Trac #15338).

From OFTC/#debian-ports:

Jul 15 2018, 7:34 AM
jrtc27 added a comment to D4959: driver: don't enable -Bsymbolic on unregisterised targets (Trac #15338).
In D4959#136709, @trofi wrote:

Anyway, we can either not do -Bsymbolic or avoid R_COPY. Perhaps it isn't possible to avoid R_COPY if we're compiling unregisterised.

Yeah. In theory -Bsymbolic can also be used on PIE binaries to avoid R_COPY but PIE does not work everywhere. I think avoiding -Bsymbolic is not too bad.

Jul 15 2018, 7:34 AM

Apr 30 2018

jrtc27 added inline comments to D4633: Allow CmmLabelDiffOff with different widths.
Apr 30 2018, 10:52 AM

Mar 9 2018

jrtc27 added a comment to D4489: SPARC nativeGen: Support for MO_SS_Conv_W32_W64.

Right, testing SPARC has historically been tricky. I do wonder whether qemu's user-space emulation might be useful here.

Mar 9 2018, 11:07 AM

Feb 1 2018

jrtc27 added a comment to D4363: Improve accuracy of get/setAllocationCounter.

@simonmar, this appears to compile on i386 with D4362. I haven't tested it for correctness yet. I'm also not entirely sure what the status of the other 32-bit architectures is. In short, this requires MO_UU_Conv W32 W64 which appears to be poorly supported.

Feb 1 2018, 11:18 AM

Oct 24 2017

jrtc27 added a comment to D4114: Move ghc{autoconf,platform,version}.h to includes/dist; hide includes/ from stage0 compiler.

I'm still struggling to find the cause of the segfaults locally; I can't seem to do anything to make the stage2 compiler *not* crash, even if I add -Iincludes to compiler_stage1_HC_OPTS. Hopefully I can get to the bottom of this soon, as it's rather tedious...

Oct 24 2017, 1:09 PM

Oct 20 2017

jrtc27 updated the diff for D4114: Move ghc{autoconf,platform,version}.h to includes/dist; hide includes/ from stage0 compiler.

Added includes/dist to rts/package.conf.in, hopefully fixing build errors.

Oct 20 2017, 6:03 AM

Oct 19 2017

jrtc27 abandoned D3741: Stg.h: Reintroduce removed EI_/II_ for bootstrapping.

Obsoleted by D4114.

Oct 19 2017, 8:04 PM
Herald added a reviewer for D4114: Move ghc{autoconf,platform,version}.h to includes/dist; hide includes/ from stage0 compiler: austin.
Oct 19 2017, 7:38 PM

Oct 16 2017

Herald added a reviewer for D4100: updateThunk: indirectee can be tagged: austin.
Oct 16 2017, 2:26 PM
Herald added a reviewer for D4099: Untag the potential AP_STACK in stg_getApStackValzh: austin.
Oct 16 2017, 2:26 PM

Oct 15 2017

Herald added a reviewer for D4093: Flags.hsc: Peek a CBool (Word8), not a Bool (Int32): austin.
Oct 15 2017, 6:03 PM

Oct 12 2017

Herald added a reviewer for D4089: RtClosureInspect: Fix inspecting Char# on 64-bit big-endian: austin.
Oct 12 2017, 5:17 PM
jrtc27 added a comment to D4089: RtClosureInspect: Fix inspecting Char# on 64-bit big-endian.
  1. This was tested on sparc64 (with my local NCG work), not powerpc64, but it was seeing the same problem.
Oct 12 2017, 5:17 PM

Sep 18 2017

jrtc27 added a comment to D3987: Re-initialise ObjectCode's info when re-mapping/re-allocating.

Sorry for the lack of detail, @angerman requested I post the diff so I did so quickly before leaving this morning.

Sep 18 2017, 1:10 PM
jrtc27 added a comment to D3987: Re-initialise ObjectCode's info when re-mapping/re-allocating.

Note: I believe PowerPC is affected by this, but I have not actually tested that hypothesis.

Sep 18 2017, 2:41 AM
Herald added a reviewer for D3987: Re-initialise ObjectCode's info when re-mapping/re-allocating: austin.
Sep 18 2017, 2:38 AM

Jul 19 2017

jrtc27 added a comment to D3741: Stg.h: Reintroduce removed EI_/II_ for bootstrapping.

Yeah, I do think something is wrong here. The stage0 compiler should be using its own version of the header.

Jul 19 2017, 2:18 PM

Jul 15 2017

jrtc27 added inline comments to D3741: Stg.h: Reintroduce removed EI_/II_ for bootstrapping.
Jul 15 2017, 1:24 PM
Herald added a reviewer for D3741: Stg.h: Reintroduce removed EI_/II_ for bootstrapping: austin.
Jul 15 2017, 1:17 PM

May 15 2017

jrtc27 added a comment to D3589: Add support for producing position-independent executables.

The command-line flags are somewhat confusing. For GCC, you compile with -fpie/-fPIE and link with -pie, and I think there's no guarantee that code compiled with -fPIC can be statically linked into a position-independent executable, although in practice I believe it's fine. Personally, I think GHC should accept -fPIE and treat it like -fPIC (with the option to behave differently in future should it so desire, so long as it continues to work for PIE) for compilation, and use this -pie option to govern linking.

Actually, what in particular do you mean? Do you think we should have both an Opt_PIE and Opt_PIC with corresponding flags? Alternatively, do you think that -fPIC and -fPIE should both affect Opt_PIC? The problem with the latter is that things like -fPIC -fno-PIE would have rather surprising behavior (although I frankly can't imagine why anyone would expect this to work anyways).

May 15 2017, 6:15 PM
jrtc27 added a comment to D3589: Add support for producing position-independent executables.

The command-line flags are somewhat confusing. For GCC, you compile with -fpie/-fPIE and link with -pie, and I think there's no guarantee that code compiled with -fPIC can be statically linked into a position-independent executable, although in practice I believe it's fine. Personally, I think GHC should accept -fPIE and treat it like -fPIC (with the option to behave differently in future should it so desire, so long as it continues to work for PIE) for compilation, and use this -pie option to govern linking.

This sounds reasonable to me. Do you think that -pie should imply -fPIE?

May 15 2017, 6:15 PM
jrtc27 added a comment to D3589: Add support for producing position-independent executables.

The command-line flags are somewhat confusing. For GCC, you compile with -fpie/-fPIE and link with -pie, and I think there's no guarantee that code compiled with -fPIC can be statically linked into a position-independent executable, although in practice I believe it's fine. Personally, I think GHC should accept -fPIE and treat it like -fPIC (with the option to behave differently in future should it so desire, so long as it continues to work for PIE) for compilation, and use this -pie option to govern linking.

May 15 2017, 5:31 PM

Nov 27 2016

jrtc27 added a comment to D2684: Fix testsuite threading, timeout, encoding and performance issues on Windows.

Looks much better style-wise; just a few more minor things.

Nov 27 2016, 8:02 PM

Nov 26 2016

jrtc27 added a comment to D2684: Fix testsuite threading, timeout, encoding and performance issues on Windows.

I've added a lot of comments picking up on style issues, some of which weren't introduced by you but might as well be tidied up since you're editing those lines.

Nov 26 2016, 7:23 PM

Nov 1 2016

jrtc27 added a comment to D2661: Sparc*: Prevent GHC from doing unaligned accesses.

Note I filed this as Trac #12793

Nov 1 2016, 4:55 AM