(cherry picked from commit 9ff26b5a)

Fix an earlier rebase mistake.

Use unsafeWithForeignPtr when possible.

(cherry picked from commit de1b380a)

(cherry picked from commit 5813cdff)

(cherry picked from commit 33a6b81e)

GHC GitLab CI authored 4 years ago and Ben Gamari committed 4 years ago

(cherry picked from commit 10739183)

This will be needed shortly.

(cherry picked from commit ef42c578)

(cherry picked from commit bc926291)

Teach it to use unsafeWithForeignPtr where appropriate.

(cherry picked from commit e0f6293e)

(cherry picked from commit 1d3e18a1)

(cherry picked from commit 3d4585fd)

(cherry picked from commit bc44a67f)

(cherry picked from commit bc178f49)

(cherry picked from commit eb21e9ea)

(cherry picked from commit 50236ed2)

(cherry picked from commit 53fb345d)

adam authored 4 years ago and Ben Gamari committed 4 years ago

(cherry picked from commit e033dd05)

adam authored 4 years ago and Ben Gamari committed 4 years ago

(cherry picked from commit c6477639)

Shayne Fletcher authored 4 years ago and Ben Gamari committed 4 years ago

(cherry picked from commit 4a437bc1)

Fixes #18994

Co-Author: Benjamin Maurer <maurer.benjamin@gmail.com>
(cherry picked from commit 3e3555cc)

Since MR !554 (#15454) GHCi automatically enabled the flag `-fobject-code` on
any module using the UnboxedTuples or UnboxedSum extensions.

MR !1553 (#16876) allowed to inhibit the automatic compiling to object-code
of these modules by setting the `fbyte-code` flag. However, it assigned 2
different semantics to this flag and introduced the regression described in
issue #18955.

This MR fixes this regression by unsetting the internal flag
`Opt_ByteCodeIfUnboxed` before it's copied to DynFlags local to the module.

This otherwise fails on newer Clangs, which warn
more aggressively on undeclared symbols.

(cherry picked from commit 3d7db148)

Strangely I find that on Alpine (and apparently only on Alpine) the
latex makeindex command expects to be given a filename, lest it reads
from stdin.

(cherry picked from commit 389a6683)

Previously we would allocate a linked list cell for each foreign export.
Now we can avoid this by taking advantage of the fact that they are
already broken into groups.

(cherry picked from commit 40dc9106)

This avoids calling `libc` in the initializers which are responsible for
registering foreign exports. We believe this should avoid the corruption
observed in #18548.

See Note [Tracking foreign exports] in rts/ForeignExports.c for an
overview of the new scheme.

(cherry picked from commit c4921349)

GHC GitLab CI authored 4 years ago and Ben Gamari committed 4 years ago

When threadPaused blackholes a thunk it calls `OVERWRITING_CLOSURE` to
zero the slop for the benefit of the sanity checker. Previously this was
done *before* pushing the thunk's free variables to the update
remembered set. Consequently we would pull zero'd pointers to the update
remembered set.

(cherry picked from commit 3e75b0db)

GHC GitLab CI authored 4 years ago and Ben Gamari committed 4 years ago

This uses the highMemDynamic flag introduced earlier to verify that
dynamic objects are properly unloaded.

(cherry picked from commit e9e1b2e7)

GHC GitLab CI authored 4 years ago and Ben Gamari committed 4 years ago

(cherry picked from commit 7a65f9e1)

Ray Shih authored 4 years ago and Ben Gamari committed 4 years ago

(This change is originally written by niteria)

This adds two functions:
* `loadNativeObj`
* `unloadNativeObj`
and implements them for Linux.

They are useful if you want to load a shared object with Haskell code
using the system linker and have GHC call dlclose() after the
code is no longer referenced from the heap.

Using the system linker allows you to load the shared object
above outside the low-mem region. It also loads the DWARF sections
in a way that `perf` understands.

`dl_iterate_phdr` is what makes this implementation Linux specific.

(cherry picked from commit 2782487f)

Fixes #16525 by tracking dependencies between object file symbols and
marking symbol liveness during garbage collection

See Note [Object unloading] in CheckUnload.c for details.

(cherry picked from commit c34a4b98)

Pulled out of !4310.

Namely, those marked with the ~"LLVM backend" label

(cherry picked from commit 7c03cc50)

As noted in #18991, we would previously allocate heap in low memory.
Due to this the linker, which typically *needs* low memory, would end up
competing with the heap. In longer builds we end up running out of
low memory entirely, leading to linking failures.

(cherry picked from commit a1a75aa9)

In ticket #18733 we noticed a rather serious deficiency in the current
fingerprinting logic for recursive groups. I have described the old
fingerprinting story and its problems in Note [Fingerprinting recursive
groups] and have reworked the story accordingly to avoid these issues.

Fixes #18733.

(cherry picked from commit 5353fd50)

(cherry picked from commit 787e93ae)