gcc/common: add disableGdbPlugin option #216237
Conversation
ghost
requested review from
trofi
and removed request for
matthewbauer,
Ericson2314,
vcunat and
Synthetica9
Sorry about that, fixed now. |
ghost
marked this pull request as draft
ofborg
bot
added
the
2.status: merge conflict
ghost
changed the title
ghost
marked this pull request as ready for review
ofborg
bot
removed
2.status: merge conflict
ofborg
bot
added
10.rebuild-darwin: 11-100
10.rebuild-linux: 11-100
and removed
10.rebuild-darwin: 501+
10.rebuild-darwin: 5001+
labels
There was a problem hiding this comment.
Apart from one flag inversion the change looks good.
Can you add a 1-sentence Why motivation description as well? A few reasons I could make up are:
- cut down on build times (does it save a lot?)
- does the default value cause some other problems during bootstrap? (which? extra depends? something probes it? some c++-specific failures?)
- just a handy switch to flip to debug some gcc aspect?
- something else?
ofborg
bot
added
10.rebuild-darwin: 0
|
Squashed, with a more detailed commmit message explaining the motivation. This raises the question of whether or not the behavior which causes us to need Also, since libcc1 is considered unmaintained there would have to be a decision about whether the benefit of a fix upstream is worth the risk of breaking things that nobody is particularly motivated to fix. |
ghost
requested review from
trofi
and removed request for
Synthetica9,
Ericson2314 and
vcunat
There was a problem hiding this comment.
Ah, early linking of libstdc++ makes sense. Yeah, disabling it early if needed is the simplest.
Generally, I saw the --with-stage1-libs= flag that might allow a similar kind of tweaking. But it's probably more fragile as it needs an exhaustive (AFAIU) list of libraries upfront.
| @@ -26,6 +27,7 @@ | |||
| , disableBootstrap ? stdenv.targetPlatform != stdenv.hostPlatform | |||
| }: | |||
|
|
|||
| assert disableGdbPlugin -> enablePlugin; | |||
There was a problem hiding this comment.
Worth removing added assert? It should be fine to disable both.
There was a problem hiding this comment.
Dammit, that should be !enablePlugin. Thank you for noticing.
There was a problem hiding this comment.
Dammit, that should be !enablePlugin. Thank you for noticing.
This commit adds an option `disableGdbPlugin` which controls whether or not the plugin *for* GDB will be built. This plugin contains a copy of `gcc`. The configure flag that this option controls is called `--disable-libcc1`. This flag name is slightly confusing: it is used only by GDB (and apparently unmaintained), yet the flag name does not mention GDB. This is why the option name is different from the configure flag name. The primary motivation for this commit is to allow the following PR (which is not yet merged) to pass `--disable-libcc1` when building the final native (build=host=target) compiler as part of the stdenv bootstrap: #209870 We need to `--disable-libcc1` in this scenario because gcc's build machinery links `libcc1` against the `libstdc++` that is part of the *compiler used to compile gcc*, rather than against the `libstdc++` that is built *by* gcc. In an FHS distribution this distinction is not terribly important because dynamically linked libraries are late-bound (ld.so resolution). However in nixpkgs this causes a reference back to the bootstrapFiles to leak all the way through to the final stdenv. More details can be found in the comment in `pkgs/stdenv/linux/default.nix` of the PR linked above. Co-authored-by: Sandro <[email protected]>
ghost
requested review from
trofi and
SuperSandro2000
and removed request for
trofi and
SuperSandro2000
|
Dammit, I messed this up again: |
ghost
mentioned this pull request
|
Caused #342904 |
Description of changes
gcc/common: add disableGdbPlugin option
This commit adds an option
disableGdbPluginwhich controls whetheror not the plugin for GDB, which contains a copy of
gcc, will bebuilt.
The configure flag that this option controls is called
--disable-libcc1. This is slightly confusing, because its is usedonly by GDB (and apparently unmaintained), yet its name does not
mention GDB. This is why the option name is different from the
configure flag name.
Things done
sandbox = trueset innix.conf? (See Nix manual)nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)