Skip to content

[pull] master from libressl-portable:master - #18

Open
pull[bot] wants to merge 1561 commits into
bazelregistry:masterfrom
libressl:master
Open

[pull] master from libressl-portable:master#18
pull[bot] wants to merge 1561 commits into
bazelregistry:masterfrom
libressl:master

Conversation

@pull

@pull pull Bot commented Nov 8, 2021

Copy link
Copy Markdown

See Commits and Changes for more details.


Created by pull[bot]

Can you help keep this open source service alive? 💖 Please sponsor : )

botovq and others added 26 commits January 26, 2026 07:22
Undoes the merge of #1211, which is no longer needed as @vszakats
confirmed by testing in the curl CI.

This reverts commit fad962e.
This patch changes to:

- add `--no-single-branch` to git clone to ensure all remote branches
  are tracked, allowing the script to switch between OpenBSD versions.

- use origin/ prefix for $openbsd_branch to fix revision lookup
  At initial setup, the branch only exists as a "remote-tracking"
  reference. Since a local branch hasn't been created yet, git cannot
  find it by name alone. Adding the origin/ prefix explicitly tells
  git to look into the remote data we just fetched.

This ensures the script can correctly identify the commit hash based
on the release timestamp even on a fresh shallow clone.

Fix #1233
MSVC does not define ssize_t.
Hence, this patch adds a typedef for compatibility
when LIBRESSL_INTERNAL is not defined.

Fix #1232
If open() fails, return the error value immediately. Adding the high bit
to -1 results in a large positive integer, which causes callers like
tls_config_load_file() to bypass their error checks.

Fix #1239
nak3 and others added 30 commits August 13, 2026 16:59
Recognize ELF riscv64 as an assembly-capable target so the
architecture-specific BN helpers are used unless assembly is
explicitly disabled.
Run CTest with parallel jobs in the Windows, Emscripten, Fedora,
and shared test-script paths to reduce test time.
per zizmor, actions/checkout should be followed by removal of credentials that
may have been saved to disk. Doing so fixes lots of whining such as:

```
warning[artipacked]: credential persistence through GitHub Actions artifacts
  --> ./.github/workflows/android.yml:27:9
   |
27 |         - name: "Checkout repository"
   |  _________^
28 | |         uses: actions/checkout@3d3c42e # v7.0.1
   | |________________________________________________________________________________^ does not set persist-credentials: false
   |
   = note: audit confidence → Low
   = note: this finding has an auto-fix
```
AIX has syslog but not vsyslog; the fallback lived in syslog.h as non-static function definitions, so it multiply-defined vsyslog_r (also defined in syslog_r.c) and pulled vasprintf in without stdio.h. Detect vsyslog and, when absent, format with vasprintf and call syslog directly from the shim.
Explicitly enable assembly for the MIPS32 and MIPS64 test jobs so that
the existing AES, BN, and SHA implementations are exercised by make
check.

Also fix the MIPS64 build to reference the generated MIPS64 assembly
sources instead of the MIPS32.
Keep assembly disabled by default on MIPS32 and MIPS64, but preserve an
explicit `--enable-asm` setting. Previously, the architecture detection
overwrote the command-line option with `enable_asm=no`.
wasm-ld only warns about unknown -z values, so the -Wl,-z,relro/now
link probes false-positive under emcc and then warn on every link.
The stack protector probe also passes (the probe program has no
protected frame) but real objects fail to link since Emscripten's
libc provides no __stack_chk_guard/__stack_chk_fail; it has its own
stack checks. -fcf-protection already fails its probe cleanly there.
Use the little endian MIPS32 toolchain since the generated assembly is
little-endian, and define the architecture feature macros needed to
select the AES, BN, SHA-1, and SHA-256 assembly implementations.

Do not generate or build SHA-512 assembly for MIPS32. The generator
accepts the O32 ABI, but the resulting implementation still uses 64-bit
instructions and requires the MIPS64 ISA. It is therefore not valid for
generic MIPS32 targets.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants