This means --enable-bootstrap now is the same as what
--enable-bootstrap=install currently does. --enable-bootstrap=install is at the
same time desupported.
We also remove --enable-bootstrap=update, which is not worth supporting.
Differential Revision: https://phabricator.services.mozilla.com/D105719
This changes things such that setting e.g. NASM=/usr/bin/nasm will avoid
bootstrapping nasm even when bootstrapping is enabled.
This is not applied to CC/CXX/HOST_CC/HOST_CXX because things are more
complicated.
This also simplifies how check_prog is called for a bootstrapped tool,
and avoids the repetition of when.
CBINDGEN handling needs the pattern being applied manually because it
currently doesn't use check_prog. Once --enable-bootstrap=install
becomes the default on developer builds, it will be possible to simplify
this.
Differential Revision: https://phabricator.services.mozilla.com/D105718
Because --with-ccache without a value fulfills a @depends_if, we need to
check for the length. And because we check the length, we can just use a
@depends.
Differential Revision: https://phabricator.services.mozilla.com/D106084
- Only bootstrap sccache when building with --with-ccache=sccache or
CCACHE=sccache
- Don't bootstrap dump_syms or nasm on builds that don't compile
Differential Revision: https://phabricator.services.mozilla.com/D105704
Instead of preemptively check for it, and then check if it's good enough to
build AV1, only check for (and bootstrap) nasm when building AV1 requires
it.
At the same time, we future-proof the code to be able to handle multiple
things requiring nasm, which we're going to add shortly.
Differential Revision: https://phabricator.services.mozilla.com/D105425
We want to find the full path to the correct `lldb-server` in the NDK.
We reference this variable in a later patch when preparing the device for
debugging.
Differential Revision: https://phabricator.services.mozilla.com/D94380
None of HAVE_NASM, HAVE_YASM, NASM_MAJOR_VERSION and NASM_MINOR_VERSION are
used. Also, the YASM variable is not necessary for old-configure anymore.
Differential Revision: https://phabricator.services.mozilla.com/D105269
* Puts the docs in order, so that contributors aren't jumping to the
middle of the page to install system tools, then back to the top to
clone Firefox.
* Removes docs on MacPorts since it's being removed in bug 1688263.
* Removes step to manually install brew packages since that happens
automatically in bootstrap now.
* Simplifies mercurial installation docs
* Removes unnecessary mozconfig-tweaking instructions
* Removes almost-always-unnecessary DEFINE and troubleshooting
information.
Differential Revision: https://phabricator.services.mozilla.com/D102973
Bug 1690930 added sysroots that can be bootstrapped. With this change,
we allow --enable-bootstrap=install to pull the right sysroot for the
configured target, and --enable-bootstrap to update it if it was already
there.
Differential Revision: https://phabricator.services.mozilla.com/D104797
We used to only push to opt-level=2 on --enable-release builds, to make
local builds faster with opt-level=1. Years later, it seems opt-level=2 makes no noticeable
difference in build times vs. opt-level=1, neither on my Threadripper
workstation at -j64 or my M1 Macbook Air at -j4.
That's one less difference to carry.
Differential Revision: https://phabricator.services.mozilla.com/D103266
2021-01-22 Kevin Jacobs <kjacobs@mozilla.com>
* automation/abi-check/previous-nss-release, lib/nss/nss.h,
lib/softoken/softkver.h, lib/util/nssutil.h:
Set version numbers to 3.62 Beta
[680ec01577b9]
2021-01-23 Kevin Jacobs <kjacobs@mozilla.com>
* tests/chains/scenarios/nameconstraints.cfg,
tests/libpkix/certs/NameConstraints.ipaca.cert,
tests/libpkix/certs/NameConstraints.ocsp1.cert:
Bug 1686134 - Renew two chains libpkix test certificates. r=rrelyea
[3ddcd845704c]
2021-01-25 Kevin Jacobs <kjacobs@mozilla.com>
* gtests/common/testvectors/hpke-vectors.h,
gtests/pk11_gtest/pk11_hpke_unittest.cc, lib/pk11wrap/pk11hpke.c,
lib/pk11wrap/pk11hpke.h, lib/pk11wrap/pk11pub.h:
Bug 1678398 - Update HPKE to draft-07. r=mt
This patch updates HPKE to draft-07. A few other minor changes are
included:
- Refactor HPKE gtests for increased parameterized testing.
- Replace memcpy calls with PORT_Memcpy
- Serialization tweaks to make way for context Export/Import (D99277).
This should not be landed without an ECH update, as fixed ECH test
vectors will otherwise fail to decrypt.
[e0bf8cadadc7]
* automation/abi-check/expected-report-libnss3.so.txt,
gtests/pk11_gtest/pk11_hpke_unittest.cc, lib/nss/nss.def,
lib/pk11wrap/pk11hpke.c, lib/pk11wrap/pk11pub.h:
Bug 1678398 - Add Export/Import functions for HPKE context. r=mt
This patch adds and exports two new HPKE functions:
`PK11_HPKE_ExportContext` and `PK11_HPKE_ImportContext`, which are
used to export a serialized HPKE context, then later reimport that
context and resume Open and Export operations. Only receiver
contexts are currently supported for export (see the rationale in
pk11pub.h).
One other change introduced here is that `PK11_HPKE_GetEncapPubKey`
now works as expected on the receiver side.
If the `wrapKey` argument is provided to the Export/Import
functions, then the symmetric keys are wrapped with AES Key Wrap
with Padding (SP800-38F, 6.3) prior to serialization.
[8bcd12ab3b34]
* automation/abi-check/expected-report-libssl3.so.txt,
gtests/ssl_gtest/libssl_internals.c,
gtests/ssl_gtest/libssl_internals.h,
gtests/ssl_gtest/ssl_extension_unittest.cc,
gtests/ssl_gtest/tls_ech_unittest.cc, lib/ssl/ssl3con.c,
lib/ssl/ssl3ext.c, lib/ssl/ssl3ext.h, lib/ssl/sslexp.h,
lib/ssl/sslimpl.h, lib/ssl/sslsecur.c, lib/ssl/sslsock.c,
lib/ssl/sslt.h, lib/ssl/tls13con.c, lib/ssl/tls13con.h,
lib/ssl/tls13ech.c, lib/ssl/tls13ech.h, lib/ssl/tls13exthandle.c,
lib/ssl/tls13exthandle.h, lib/ssl/tls13hashstate.c,
lib/ssl/tls13hashstate.h:
Bug 1681585 - Update ECH to Draft-09. r=mt
This patch updates ECH implementation to draft-09. Changes of note
are:
- Acceptance signal derivation is now based on the handshake secret.
- `config_id` hint changes from 32B to 8B, trial decryption added on
the server.
- Duplicate code in HRR cookie handling has been consolidated into
`tls13_HandleHrrCookie`.
- `ech_is_inner` extension is added, which causes a server to indicate
ECH acceptance.
- Per the above, support signaling ECH acceptance when acting as a
backend server in split-mode (i.e. when there is no other local
Encrypted Client Hello state).
[ed07a2e2a124]
2021-01-24 Kevin Jacobs <kjacobs@mozilla.com>
* cmd/selfserv/selfserv.c:
Bug 1681585 - Add ECH support to selfserv. r=mt
Usage example: mkdir dbdir && cd dbdir certutil -N -d . certutil -S
-s "CN=ech-public.com" -n ech-public.com -x -t "C,C,C" -m 1234 -d .
certutil -S -s "CN=ech-private-backend.com" -n ech-private-
backend.com -x -t "C,C,C" -m 2345 -d . ../dist/Debug/bin/selfserv -a
ech-public.com -a ech-private-backend.com -n ech-public.com -n ech-
private-backend.com -p 8443 -d dbdir/ -X publicname:ech-public.com
(Copy echconfig from selfserv output and paste into the below
command) ../dist/Debug/bin/tstclnt -D -p 8443 -v -A
tests/ssl/sslreq.dat -h ech-private-backend.com -o -N <echconfig> -v
[92dcda94c1d4]
Differential Revision: https://phabricator.services.mozilla.com/D102982
This adds a --enable-bootstrap build flag that will automatically update
cbindgen, node, clang, sccache, nasm, wine, lucetc, dump_syms, pdbstr,
and winchecksec if they are already installed in ~/.mozbuild.
Eventually, we'll want to allow to install toolchains that weren't
already install, but one step at a time.
This explicitly doesn't cover rustc, which is its own can of worms, or
android-{ndk,sdk}, which are not installed via toolchain artifacts
currently.
Differential Revision: https://phabricator.services.mozilla.com/D101723
Instead of adding all possible tool paths from ~/.mozbuild, we only
add the relevant paths for each of the tools we search for.
Differential Revision: https://phabricator.services.mozilla.com/D101718
The only thing that varies between toolchain_search_path and
host_toolchain_search_path is the path to the MSVC C/C++ compiler and
tools, because MSVC has a different compiler for each platform, and host
and target platforms may differ (when e.g. compiling for arm64 on
x86_64).
However, we don't use the MSVC compiler anymore, and the only thing we
use its path for is the assembler, which we don't use for host things
(and we don't have a HOST_AS), and to derive the path to some system
headers/SDK.
Differential Revision: https://phabricator.services.mozilla.com/D101714
There is only one place where it's used:
config/check_vanilla_allocations.py, which is only executed from
js/src/build/Makefile.in on the condition that the build is targeting
Linux and not LTO. But the LTO test is actually outdated, because we
don't build with `-flto`, but `-flto=thin`, so the exclusion doesn't
work anymore.
There is however no AC_CHECK_PROG, and we currently rely on NM to be
given, or fall back to "nm", which works in most cases, except LTO with
clang. It works on CI because in LTO builds we explicitly set NM to
llvm-nm (which can output symbols from LLVM bitcode objects), but we
could also do that automatically.
So we add a full detection of nm/llvm-nm to python configure, and limit
it to Linux, since we only ever use it there.
Differential Revision: https://phabricator.services.mozilla.com/D101681
js/src/aclocal.m4 contains includes starting with `../../`.
As explained in https://bugzilla.mozilla.org/show_bug.cgi?id=1680862#c8,
m4 will first try to resolve this path relative to the working directory
and only if that path doesn't exists, fall back to the location set by
`localdir` (from `-I`).
The working directory is usually MOZ_OBJDIR, an immediate subdirectory
of topsrcdir, so `../../` resolves to a location outside of topsrcdir.
Usually, that path does not exist, and m4 falls back to `localdir` that
was passed via `-I`.
But if that path existed and is incompatible with the current Gecko
checkout, then the build will fail (see bug report). To prevent this
from happening, this patch fixes the working directory to `localdir`,
so that m4 will immediately find the expected file.
Differential Revision: https://phabricator.services.mozilla.com/D101500
In addition to warning on regular methods overloading virtual functions,
GCC also watches for static functions doing such overloads.
:andi confirmed that this is not valuable, so the warning is being
disabled for GCC.
Differential Revision: https://phabricator.services.mozilla.com/D101367
Passing `-flto=thin` worked previously but the value passed was just ignored
and full lto was performed. On newer versions of gcc passing an unknown value
causes failure. So this commit checks if `-flto=thin` is passed and fails with
an error message if so, else full lto is enabled if any other value is passed.
Differential Revision: https://phabricator.services.mozilla.com/D100953
Some distros include flags when they specify the location of a binary,
such as: `XARGS=xargs -r`.
This was confusing in `configure`, since:
* We require that environment variables contain only path
overrides (without flags).
* The error message thrown when configure chokes was unclear: "why
would `$ xargs -r` fail?"
This patch should make our "path-only" requirement more clear.
Differential Revision: https://phabricator.services.mozilla.com/D100044
Old clang shakes its fist when `auto&& item : range` is used with a
range
that returns values instead of references.
Modern `clang` doesn't warn for this scenario, so we disable the
warning.
Also removes pragmas that manually disable this warning.
Differential Revision: https://phabricator.services.mozilla.com/D100155
I was waiting for a better reason to do this, because the cbindgen
changes from 0.15.0 to 0.16.0 don't break trunk builds. But since
downstream has updated (see bug 1684180) and there's no reason not to,
let's do this to avoid future churn.
Differential Revision: https://phabricator.services.mozilla.com/D100499
2020-12-11 Kevin Jacobs <kjacobs@mozilla.com>
* automation/abi-check/expected-report-libssl3.so.txt, automation/abi-
check/previous-nss-release, lib/nss/nss.h, lib/softoken/softkver.h,
lib/util/nssutil.h:
Set version numbers to 3.61 Beta
[f277d2674c80]
* gtests/<...>
Bug 1677207 - Update Google Test to release-1.10.0 r=bbeurdouche
./gtests/google_test/update.sh release-1.10.0 && hg remove -A && hg
add gtests/google_test/*
[89141382df45]
* gtests/<...>
Bug 1677207 - Replace references to TestCase, which is deprecated,
with TestSuite r=bbeurdouche
grep -rl --exclude-dir=google_test INSTANTIATE_TEST_CASE_P gtests |
xargs sed -i '' s/INSTANTIATE_TEST_CASE_P/INSTANTIATE_TEST_SUITE_P/g
grep -rl --exclude-dir=google_test SetUpTestCase gtests | xargs sed
-i '' s/SetUpTestCase/SetUpTestSuite/g
[e15b78be87fa]
* gtests/ssl_gtest/ssl_ciphersuite_unittest.cc,
gtests/ssl_gtest/ssl_debug_env_unittest.cc,
gtests/ssl_gtest/ssl_extension_unittest.cc,
gtests/ssl_gtest/ssl_loopback_unittest.cc,
gtests/ssl_gtest/ssl_renegotiation_unittest.cc,
gtests/ssl_gtest/ssl_resumption_unittest.cc,
gtests/ssl_gtest/ssl_version_unittest.cc,
gtests/ssl_gtest/tls_ech_unittest.cc:
Bug 1677207 - Use GTEST_SKIP in ssl_gtests. r=bbeurdouche
[0772f1bf5fd6]
2020-12-17 Robert Relyea <rrelyea@redhat.com>
* gtests/common/testvectors/ike-aesxcbc-vectors.h,
gtests/common/testvectors/ike-sha1-vectors.h,
gtests/common/testvectors/ike-sha256-vectors.h,
gtests/common/testvectors/ike-sha384-vectors.h,
gtests/common/testvectors/ike-sha512-vectors.h,
gtests/common/testvectors_base/test-structs.h,
gtests/pk11_gtest/manifest.mn, gtests/pk11_gtest/pk11_gtest.gyp,
gtests/pk11_gtest/pk11_ike_unittest.cc, lib/softoken/sftkike.c:
Bug 1682071 IKE Quick mode IPSEC give you incorrect keys if you are
asking for keys smaller than the hash size.
IKE Appendix B fixes.
This patch fixes 2 problems.
If you run either ike v1 App B or quick mode asking for a key with
length
mod macsize = 0, you will generate an extra block that's not used
and overwrites the end of the buffer.
If you use quick mode, the function incorrectly subsets the
existing key
rather than generating a new key. This is correct behavior for
Appendix B, where appendix B is trying to take a generated key and
create a new longer key (with no diversification, just transform the
key into something that's longer), so if you ask for a key less than
or equal to, then you want to just subset the original key. In quick
mode you are taking a base key and creating a set of new keys based
on additional data, so you want to subset the generated data. This
patch only subsets the original key if you aren't doing quickmode.
Full test vectors have now been added for all ike modes in this
patch as well (previously we depended on the FIPS CAVS tests to test
ike, which covers basic IKEv1, IKEv1_psk, and IKEv2 but not IKEv1
App B and IKE v1 Quick mode).
[f4995c9fa185]
2020-12-18 Robert Relyea <rrelyea@redhat.com>
* gtests/common/testvectors/rsa_pkcs1_2048_test-vectors.h,
gtests/common/testvectors/rsa_pkcs1_3072_test-vectors.h,
gtests/common/testvectors/rsa_pkcs1_4096_test-vectors.h,
gtests/freebl_gtest/Makefile, gtests/freebl_gtest/manifest.mn,
gtests/freebl_gtest/rsa_unittest.cc, gtests/manifest.mn,
gtests/pk11_gtest/pk11_rsaencrypt_unittest.cc,
gtests/pk11_gtest/pk11_rsaoaep_unittest.cc, lib/freebl/alghmac.c,
lib/freebl/alghmac.h, lib/freebl/rsapkcs.c:
Bug 1651411 New tlsfuzzer code can still detect timing issues in RSA
operations.
This patch defeats Bleichenbacher by not trying to hide the size of
the decrypted text, but to hide if the text succeeded for failed.
This is done by generating a fake returned text that's based on the
key and the cipher text, so the fake data is always the same for the
same key and cipher text. Both the length and the plain text are
generated with a prf.
Here's the proposed spec the patch codes to:
1. Use SHA-256 to hash the private exponent encoded as a big-
endian integer to a string the same length as the public modulus.
Keep this value secret. (this is just an optimisation so that the
implementation doesn't have to serialise the key over and over
again) 2. Check the length of input according to step one of
https://tools.ietf.org/html/rfc8017#section-7.2.2 3. When provided
with a ciphertext, use SHA-256 HMAC(key=hash_from_step1,
text=ciphertext) to generate the key derivation key 4. Use SHA-256
HMAC with key derivation key as the key and a two-byte big- endian
iterator concatenated with byte string "length" with the big- endian
representation of 2048 (0x0800) as the bit length of the generated
string.
- Iterate this PRF 8 times to generate a 256 byte string 5. initialise
the length of synthetic message to 0 6. split the PRF output into 2
byte strings, convert into big-endian integers, zero- out high-order
bits so that they have the same bit length as the octet length of
the maximum acceptable message size (k-11), select the last integer
that is no larger than (k-11) or remain at 0 if no integer is
smaller than (k-11); this selection needs to be performed using a
side-channel free operators 7. Use SHA-256 HMAC with key derivation
key as the key and a two-byte big-endian iterator concatenated with
byte string "message" with the big-endian representation of k*8
- use this PRF to generate k bytes of output (right-truncate last HMAC
call if the number of generated bytes is not a multiple of SHA-256
output size) 8. perform the RSA decryption as described in step 2 of
section 7.2.2 of rfc8017 9. Verify the EM message padding as
described in step 3 of section 7.2.2 of rfc8017, but instead of
outputting "decryption error", return the last l bytes of the
"message" PRF, when l is the selected synthetic message length using
the "length" PRF, make this decision and copy using side-channel
free operation
[fc05574c7399]
2020-12-22 Robert Relyea <rrelyea@redhat.com>
* gtests/freebl_gtest/rsa_unittest.cc,
gtests/pk11_gtest/pk11_rsaoaep_unittest.cc, lib/freebl/alghmac.c,
lib/freebl/rsapkcs.c:
Restore lost portion of the bleichenbacher timing batch that
addressed review comments. All the review comments pertained to
actual code comments, so this patch only affects the comments.
[fcebe146314e]
2020-12-22 Kevin Jacobs <kjacobs@mozilla.com>
* lib/dev/devslot.c:
Bug 1682863 - Revert nssSlot_IsTokenPresent to 3.58 after ongoing Fx
hangs with slow PKCS11 devices. r=bbeurdouche
This patch reverts the `nssSlot_IsTokenPresent` changes made in bug
1663661 and bug 1679290, restoring the version used in NSS 3.58 and
earlier. It's not an actual `hg backout` because the comment in
lib/dev/devt.h is worth keeping. While removing the nested locking
did resolve the hang for some (most?) third-party modules, problems
remain with some slower tokens after an even further relaxation of
the locking, which defeats the purpose of addressing the races in
the first place.
The crash addressed by these patches was caused by the Intermediate
Preloading Healer in Firefox, which has been disabled. We clearly
have insufficient test coverage for third-party modules, and now
that osclientcerts is enabled in Fx Nightly, any problems caused by
these and similar changes is unlikely to be reported until Fx Beta,
well after NSS RTM. I think the best option at this point is to
simply revert NSS.
[97ef009f7a78] [tip]
Differential Revision: https://phabricator.services.mozilla.com/D100401
The gyp flag logic in nICEr is supposed to ensure that the code is instrumented
for libFuzzer because we have a related fuzzing target. However, libFuzzer
instrumentation must be completely disabled for TSan due to incompatibility.
The current logic fails in doing so and incorrectly falls back to legacy
trace-pc instrumentation causing the TSan fuzzing build to fail on startup.
Differential Revision: https://phabricator.services.mozilla.com/D99351
This is bug 1639318 all over again except from Rust rather than C++. It's the same symptom, nsXPTCStub vtables aren't marked as valid targets.
In the earlier bug we disabled CFG for C++ on ARM64. Let's do the same for Rust. According to that bug, "It's not clear why this doesn't happen on x86 builds. Given priorities, I can't really justify investigating this, although I suspect that fixing the underlying issue would be pretty much bug 1483885."
Differential Revision: https://phabricator.services.mozilla.com/D99278
Bug 1680152 updated automation to use the 10.12 SDK, and shortly after,
bug 1678174 introduced a change that broke the build with the 10.11 SDK.
Considering we haven't actually supported running on macos 10.11 and
earlier since Firefox 79, and that still supporting building with the
10.11 SDK would mean adding and maintaining code that, in practice,
would never be used by users, I think it is fair at this point that we
just drop support for the 10.11 SDK entirely.
Differential Revision: https://phabricator.services.mozilla.com/D99181
On a very parallel debug build, I see a long time just waiting for
bindgen / style compilation / geckoservo.
Turns out that a bunch of this is just proc macros / build scripts.
Optimizing it saves between 10 and 17 seconds of my debug build. We
might want to consider running bindgen much like cbindgen rather than
rebuilding it all the time, which should help a lot more, but my guess
is that this should still help with the pretty hot custom derives that
the style crate runs.
This needs rust 1.41, so the requirement for tools/crashreporter needs
to be bumped as a consequence. To make things simpler, it was bumped
to 1.47 while we're at it.
Differential Revision: https://phabricator.services.mozilla.com/D98366
Windows being Windows, with its own spellings for everything, let's stay within the WINNT block regardless of whether we meet the compiler requirement.
Differential Revision: https://phabricator.services.mozilla.com/D98355
Code review followups for 1675600: Restrict the use of new pass manager during LTO to builds where we're using the new pass manager in general, and (on Windows) where lld-link is new enough to understand the flag.
Differential Revision: https://phabricator.services.mozilla.com/D97372
Upstream clang supports --target=aarch64-apple-darwin, and that matches
what config.sub canonicalizes to, and thus what the default toolchain
prefix ends up being, so it's better to use aarch64-apple-darwin when
not compiling with Xcode clang.
Differential Revision: https://phabricator.services.mozilla.com/D97698
There is no earlier SDK that supports it. It seems Xcode clang doesn't
care (maybe it defaults to 11.0 is MACOSX_DEPLOYMENT_TARGET is too low?),
but upstream clang does.
Differential Revision: https://phabricator.services.mozilla.com/D97565
And don't set it via mozconfig. The default to /System/Library/PrivateFrameworks
may also not have matched the used SDK previously, so the new default is
better.
Differential Revision: https://phabricator.services.mozilla.com/D97564
2020-11-18 Kevin Jacobs <kjacobs@mozilla.com>
* lib/ssl/ssl3con.c, lib/ssl/tls13con.c, lib/ssl/tls13ech.c:
Bug 1654332 - Fixup a10493dcfcc9: copy ECHConfig.config_id with
socket r=jcj
A late review change for ECH was for the server to compute each
ECHConfig `config_id` when set to the socket, rather than on each
connection. This works, but now we also need to copy that config_id
when copying a socket, else the server won't find a matching
ECHConfig to use for decryption.
[3eacb92e9adf] [tip]
2020-11-17 Kevin Jacobs <kjacobs@mozilla.com>
* automation/abi-check/expected-report-libssl3.so.txt,
cmd/tstclnt/tstclnt.c, cpputil/tls_parser.h,
gtests/ssl_gtest/libssl_internals.c,
gtests/ssl_gtest/libssl_internals.h, gtests/ssl_gtest/manifest.mn,
gtests/ssl_gtest/ssl_auth_unittest.cc,
gtests/ssl_gtest/ssl_custext_unittest.cc,
gtests/ssl_gtest/ssl_extension_unittest.cc,
gtests/ssl_gtest/ssl_gtest.gyp,
gtests/ssl_gtest/ssl_tls13compat_unittest.cc,
gtests/ssl_gtest/tls_agent.cc, gtests/ssl_gtest/tls_agent.h,
gtests/ssl_gtest/tls_connect.cc, gtests/ssl_gtest/tls_connect.h,
gtests/ssl_gtest/tls_ech_unittest.cc,
gtests/ssl_gtest/tls_esni_unittest.cc,
gtests/ssl_gtest/tls_filter.cc, gtests/ssl_gtest/tls_filter.h,
lib/ssl/SSLerrs.h, lib/ssl/manifest.mn, lib/ssl/ssl.gyp,
lib/ssl/ssl3con.c, lib/ssl/ssl3ext.c, lib/ssl/ssl3ext.h,
lib/ssl/ssl3exthandle.c, lib/ssl/ssl3exthandle.h,
lib/ssl/ssl3prot.h, lib/ssl/sslencode.c, lib/ssl/sslencode.h,
lib/ssl/sslerr.h, lib/ssl/sslexp.h, lib/ssl/sslimpl.h,
lib/ssl/sslinfo.c, lib/ssl/sslsecur.c, lib/ssl/sslsock.c,
lib/ssl/sslt.h, lib/ssl/tls13con.c, lib/ssl/tls13con.h,
lib/ssl/tls13ech.c, lib/ssl/tls13ech.h, lib/ssl/tls13esni.c,
lib/ssl/tls13esni.h, lib/ssl/tls13exthandle.c,
lib/ssl/tls13exthandle.h, lib/ssl/tls13hashstate.c,
lib/ssl/tls13hashstate.h:
Bug 1654332 - Update ESNI to draft-08 (ECH). r=mt
This patch adds support for Encrypted Client Hello (draft-ietf-tls-
esni-08), replacing the existing ESNI (draft -02) support.
There are five new experimental functions to enable this:
- SSL_EncodeEchConfig: Generates an encoded (not BASE64) ECHConfig
given a set of parameters.
- SSL_SetClientEchConfigs: Configures the provided ECHConfig to the
given socket. When configured, an ephemeral HPKE keypair will be
generated for the CH encryption.
- SSL_SetServerEchConfigs: Configures the provided ECHConfig and
keypair to the socket. The keypair specified will be used for HPKE
operations in order to decrypt encrypted Client Hellos as they are
received.
- SSL_GetEchRetryConfigs: If ECH is rejected by the server and
compatible retry_configs are provided, this API allows the
application to extract those retry_configs for use in a new
connection.
- SSL_EnableTls13GreaseEch: When enabled, non-ECH Client Hellos will
have a "GREASE ECH" (i.e. fake) extension appended. GREASE ECH is
disabled by default, as there are known compatibility issues that
will be addressed in a subsequent draft.
The following ESNI experimental functions are deprecated by this
update:
- SSL_EncodeESNIKeys
- SSL_EnableESNI
- SSL_SetESNIKeyPair
In order to be used, NSS must be compiled with
`NSS_ENABLE_DRAFT_HPKE` defined.
[a10493dcfcc9]
* lib/ssl/ssl3con.c, lib/ssl/sslencode.c, lib/ssl/sslencode.h,
lib/ssl/tls13con.c, lib/ssl/tls13con.h:
Bug 1654332 - Buffered ClientHello construction. r=mt
This patch refactors construction of Client Hello messages. Instead
of each component of the message being written separately into
`ss->sec.ci.sendBuf`, we now construct the message in its own
sslBuffer. Once complete, the entire message is added to the sendBuf
via `ssl3_AppendHandshake`.
`ssl3_SendServerHello` already uses this approach and it becomes
necessary for ECH, where we use the constructed ClientHello to
create an inner ClientHello.
[d40121ba59ba]
2020-11-13 J.C. Jones <jjones@mozilla.com>
* automation/abi-check/expected-report-libnss3.so.txt, automation/abi-
check/expected-report-libnssutil3.so.txt, automation/abi-check
/previous-nss-release, lib/nss/nss.h, lib/softoken/softkver.h,
lib/util/nssutil.h:
Set version numbers to 3.60 Beta
[5e7b37609f22]
Differential Revision: https://phabricator.services.mozilla.com/D97492
The SEH unwind info problem no longer happens because nowadays bug 1631929 prevents the code pattern that led to it. I've confirmed that bug 1626951's bustage doesn't come back, both in regular and beta-simulation builds.
This cleanup will allow me to reverse the dependency in bug 1677742 and have `lto` depend on `new_pass_manager_flags`.
Differential Revision: https://phabricator.services.mozilla.com/D97362
Currently, we report "10.0.0.or.more" for the latest release of Xcode,
but we know it's 10.0.0 for sure, which is when the clang version
reported by Xcode is exactly 12.0.0. We know that in the past a bump of
LLVM has always been accompanied with a bump of the minor version, so we
expect no LLVM version bump until at least 12.0.1.
Differential Revision: https://phabricator.services.mozilla.com/D97241
Currently, in LTO builds, we use the new pass manager during the initial translation to bitcode but not for the final optimization during linking.
On Linux, we can enable the new pass manager during LTO with a plugin option. I've landed a patch upstream to allow it on Windows as well, which is included here.
Switching the pass manager brings speed improvements on its own, but it also reduces code size by ~6%, which we can use a portion of as budget to increase the import limit (via the hot multiplier) for even more speed improvements.
Differential Revision: https://phabricator.services.mozilla.com/D96108
Allow-list all Python code in tree for use with the black linter, and re-format all code in-tree accordingly.
To produce this patch I did all of the following:
1. Make changes to tools/lint/black.yml to remove include: stanza and update list of source extensions.
2. Run ./mach lint --linter black --fix
3. Make some ad-hoc manual updates to python/mozbuild/mozbuild/test/configure/test_configure.py -- it has some hard-coded line numbers that the reformat breaks.
4. Make some ad-hoc manual updates to `testing/marionette/client/setup.py`, `testing/marionette/harness/setup.py`, and `testing/firefox-ui/harness/setup.py`, which have hard-coded regexes that break after the reformat.
5. Add a set of exclusions to black.yml. These will be deleted in a follow-up bug (1672023).
# ignore-this-changeset
Differential Revision: https://phabricator.services.mozilla.com/D94045
Allow-list all Python code in tree for use with the black linter, and re-format all code in-tree accordingly.
To produce this patch I did all of the following:
1. Make changes to tools/lint/black.yml to remove include: stanza and update list of source extensions.
2. Run ./mach lint --linter black --fix
3. Make some ad-hoc manual updates to python/mozbuild/mozbuild/test/configure/test_configure.py -- it has some hard-coded line numbers that the reformat breaks.
4. Make some ad-hoc manual updates to `testing/marionette/client/setup.py`, `testing/marionette/harness/setup.py`, and `testing/firefox-ui/harness/setup.py`, which have hard-coded regexes that break after the reformat.
5. Add a set of exclusions to black.yml. These will be deleted in a follow-up bug (1672023).
# ignore-this-changeset
Differential Revision: https://phabricator.services.mozilla.com/D94045
biplist broke with python 3.9, but we don't need this anymore because
Python3's plistlib allows reading binary plists since python 3.4.
I've tested this with all sdks that I have and we support (10.11-10.15).
Differential Revision: https://phabricator.services.mozilla.com/D94577
This commit also allows `memfd_create` in the seccomp-bpf policy for all
process types.
`memfd_create` is an API added in Linux 3.17 (and adopted by FreeBSD
for the upcoming version 13) for creating anonymous shared memory
not connected to any filesystem. Supporting it means that sandboxed
child processes on Linux can create shared memory directly instead of
messaging a broker, which is unavoidably slower, and it should avoid
the problems we'd been seeing with overly small `/dev/shm` in container
environments (which were causing serious problems for using Firefox for
automated testing of frontend projects).
`memfd_create` also introduces the related operation of file seals:
irrevocably preventing types of modifications to a file. Unfortunately,
the most useful one, `F_SEAL_WRITE`, can't be relied on; see the large
comment in `SharedMemory:ReadOnlyCopy` for details. So we still use
the applicable seals as defense in depth, but read-only copies are
implemented on Linux by using procfs (and see the comments on the
`ReadOnlyCopy` function in `shared_memory_posix.cc` for the subtleties
there).
There's also a FreeBSD implementation, using `cap_rights_limit` for
read-only copies, if the build host is new enough to have the
`memfd_create` function.
The support code for Android, which doesn't support shm_open and can't
use the memfd backend because of issues with its SELinux policy (see bug
1670277), has been reorganized to reflect that we'll always use its own
API, ashmem, in that case.
Differential Revision: https://phabricator.services.mozilla.com/D90605
The warning is broken in older versions of clang. It also turns out we
were disabling it locally for some directories because of this very
problem.
A few local disable rules stay under accessible/ because they do hide
actual warnings from code generated by MIDL.
Differential Revision: https://phabricator.services.mozilla.com/D94262
We used to have a complicated scheme to figure out the minimum supported
version of clang on OSX, based on some compiler feature, which wouldn't
allow to do other version checks further down the line.
The main blocker for better tests was to be able to distinguish between
Xcode clang and plain clang, which turns out to be possible with the
__apple_build_version__ define.
We still need to map versions manually, but it's better than the current
status quo.
Differential Revision: https://phabricator.services.mozilla.com/D94261
This commit also allows `memfd_create` in the seccomp-bpf policy for all
process types.
`memfd_create` is an API added in Linux 3.17 (and adopted by FreeBSD
for the upcoming version 13) for creating anonymous shared memory
not connected to any filesystem. Supporting it means that sandboxed
child processes on Linux can create shared memory directly instead of
messaging a broker, which is unavoidably slower, and it should avoid
the problems we'd been seeing with overly small `/dev/shm` in container
environments (which were causing serious problems for using Firefox for
automated testing of frontend projects).
`memfd_create` also introduces the related operation of file seals:
irrevocably preventing types of modifications to a file. Unfortunately,
the most useful one, `F_SEAL_WRITE`, can't be relied on; see the large
comment in `SharedMemory:ReadOnlyCopy` for details. So we still use
the applicable seals as defense in depth, but read-only copies are
implemented on Linux by using procfs (and see the comments on the
`ReadOnlyCopy` function in `shared_memory_posix.cc` for the subtleties
there).
There's also a FreeBSD implementation, using `cap_rights_limit` for
read-only copies, if the build host is new enough to have the
`memfd_create` function.
The support code for Android, which doesn't support shm_open and can't
use the memfd backend because of issues with its SELinux policy (see bug
1670277), has been reorganized to reflect that we'll always use its own
API, ashmem, in that case.
Differential Revision: https://phabricator.services.mozilla.com/D90605
Allow-list all Python code in tree for use with the black linter, and re-format all code in-tree accordingly.
To produce this patch I did all of the following:
1. Make changes to tools/lint/black.yml to remove include: stanza and update list of source extensions.
2. Run ./mach lint --linter black --fix
3. Make some ad-hoc manual updates to python/mozbuild/mozbuild/test/configure/test_configure.py -- it has some hard-coded line numbers that the reformat breaks.
4. Add a set of exclusions to black.yml. These will be deleted in a follow-up bug (1672023).
# ignore-this-changeset
Differential Revision: https://phabricator.services.mozilla.com/D94045
This avoids a sort of duplication of work between both, because the
linker will eventually LTO-compile everything, so we technically don't
really need the extra step of the rust compiler doing an intermediate
LTO on the static libraries it produces.
Differential Revision: https://phabricator.services.mozilla.com/D94224
2020-10-13 Mike Hommey <mh@glandium.org>
* lib/freebl/freebl.gyp:
Bug 1670839 - Use ARM crypto extension for AES, SHA1 and SHA2 on
mac. r=kjacobs
AFAICT, the Makefile equivalent already does.
[58dc3216d518] [tip]
* lib/freebl/sha1-armv8.c:
Bug 1670839 - Only build sha1-armv8.c code when USE_HW_SHA1 is
defined. r=kjacobs
This matches what is done in sha256-armv8.c, and avoids
inconsistency with sha1-fast.c, which will define the same functions
in the case USE_HW_SHA1 is not defined.
[54be084e3ba8]
2020-10-16 J.C. Jones <jjones@mozilla.com>
* automation/abi-check/expected-report-libnss3.so.txt, automation/abi-
check/previous-nss-release, lib/nss/nss.h, lib/softoken/softkver.h,
lib/util/nssutil.h:
Set version numbers to 3.59 Beta
[d4b21706e432]
Differential Revision: https://phabricator.services.mozilla.com/D94070
Rustc >= 1.44 changed the file names of the static libraries it
produces with -windows-gnu targets, to match that of mingw clang/gcc.
Considering we still build on 1.43, the best fix would be to derive the
prefix/suffix based on the version of rust, but that actually turns into
a hard-to-solve problem because of configure tests for bindgen also
depending on the prefix/suffix value to be known.
On the other hand, we're soon due to an update to 1.47, so the simpler
solution is to just push mingw builds to require 1.44 (settling for the
smallest upgrade possible for now) and to remove the split between C and
rust library prefix/suffixes.
Differential Revision: https://phabricator.services.mozilla.com/D93726
This commit also allows `memfd_create` in the seccomp-bpf policy for all
process types.
`memfd_create` is an API added in Linux 3.17 (and adopted by FreeBSD
for the upcoming version 13) for creating anonymous shared memory
not connected to any filesystem. Supporting it means that sandboxed
child processes on Linux can create shared memory directly instead of
messaging a broker, which is unavoidably slower, and it should avoid
the problems we'd been seeing with overly small `/dev/shm` in container
environments (which were causing serious problems for using Firefox for
automated testing of frontend projects).
`memfd_create` also introduces the related operation of file seals:
irrevocably preventing types of modifications to a file. Unfortunately,
the most useful one, `F_SEAL_WRITE`, can't be relied on; see the large
comment in `SharedMemory:ReadOnlyCopy` for details. So we still use
the applicable seals as defense in depth, but read-only copies are
implemented on Linux by using procfs (and see the comments on the
`ReadOnlyCopy` function in `shared_memory_posix.cc` for the subtleties
there).
There's also a FreeBSD implementation, using `cap_rights_limit` for
read-only copies, if the build host is new enough to have the
`memfd_create` function.
Differential Revision: https://phabricator.services.mozilla.com/D90605
Now that we don't recurse into the js python configure, we don't need to
have a special treatment for the options that need to be passed down to
that subconfigure, which is what js_option was for.
Differential Revision: https://phabricator.services.mozilla.com/D92727
It was only meant to be used internally, when the top-level python
configure invoked the js python subconfigure. Now that this doesn't
happen, we can remove the option, and consolidate js_standalone and
building_js, which are now roughly synonyms.
Differential Revision: https://phabricator.services.mozilla.com/D92726
Instead, we now run js/src/old-configure from the top-level configure
after having run old-configure and extracted a few variables to inherit
from it.
Because we're now running from the top-level, $_objdir is always the
top-level objdir, which simplifies some things. The topobjdir in
js/src/config.status, however, needs to stay in js/src because of the
build frontend expecting it there.
When running js/src/old-configure, we used to need some special
treatment for a large number of variables for historic reasons, where
we'd take values from the assigned values before running old-configure
for some, or from AC_SUBSTs after running old-configure.
Now that both old-configure and js/src/old-configure get the same
assignments from old-configure.vars, we don't need anything special for
the former. And only a few remaining variables still need manual work
for the latter.
One notable difference, though, is that the new code doesn't try to
avoid running js subconfigure, which added complexity, and was actually
error-prone.
Differential Revision: https://phabricator.services.mozilla.com/D92725
Preparing to run both old-configure and js/src/old-configure from the
same python configure run, we refactor things such that shared parts are
separate.
Differential Revision: https://phabricator.services.mozilla.com/D92722
It is only really used in js/src/devtools/rootAnalysis/Makefile.in,
and even there, the way it is used seems wrong, so fix that at the
same time (binaries have been linked into $DIST/bin directly for a
while).
Differential Revision: https://phabricator.services.mozilla.com/D92721
In order to be able to run both old-configure and js/src/old-configure
from the same python configure run, we need to stop setting the items
set by old-configure into the global sandbox config, and instead store
them to be later handled by configure.py.
Differential Revision: https://phabricator.services.mozilla.com/D92718
This commit also allows `memfd_create` in the seccomp-bpf policy for all
process types.
`memfd_create` is an API added in Linux 3.17 (and adopted by FreeBSD
for the upcoming version 13) for creating anonymous shared memory
not connected to any filesystem. Supporting it means that sandboxed
child processes on Linux can create shared memory directly instead of
messaging a broker, which is unavoidably slower, and it should avoid
the problems we'd been seeing with overly small `/dev/shm` in container
environments (which were causing serious problems for using Firefox for
automated testing of frontend projects).
`memfd_create` also introduces the related operation of file seals:
irrevocably preventing types of modifications to a file. Unfortunately,
the most useful one, `F_SEAL_WRITE`, can't be relied on; see the large
comment in `SharedMemory:ReadOnlyCopy` for details. So we still use
the applicable seals as defense in depth, but read-only copies are
implemented on Linux by using procfs (and see the comments on the
`ReadOnlyCopy` function in `shared_memory_posix.cc` for the subtleties
there).
There's also a FreeBSD implementation, using `cap_rights_limit` for
read-only copies, if the build host is new enough to have the
`memfd_create` function.
Differential Revision: https://phabricator.services.mozilla.com/D90605
2020-09-18 Kevin Jacobs <kjacobs@mozilla.com>
* automation/abi-check/previous-nss-release, lib/nss/nss.h,
lib/softoken/softkver.h, lib/util/nssutil.h:
Set version numbers to 3.58 Beta
[c28e20f61e5d] [tip]
* .hgtags:
Added tag NSS_3_57_RTM for changeset cf7e3e8abd77
[a963849538ca] <NSS_3_57_BRANCH>
* lib/nss/nss.h, lib/softoken/softkver.h, lib/util/nssutil.h:
Set version numbers to 3.57 final
[cf7e3e8abd77] [NSS_3_57_RTM] <NSS_3_57_BRANCH>
Differential Revision: https://phabricator.services.mozilla.com/D91070
It was added in bug 1320656 because back then we were building as C++14
with warnings about future incompatibilities with C++17. Since then,
we've switched to C++17, which means we had to fix those
incompatibilities, and thus they don't exist anymore. A local build with
-Werror=noexcept-type finishes just fine.
This removes the only difference between top-level and js warning flags.
Differential Revision: https://phabricator.services.mozilla.com/D90521
I think they're remnants from the past that we don't really need anymore.
And they're making things more complicated for some pending work of mine.
Differential Revision: https://phabricator.services.mozilla.com/D89687
I think they're remnants from the past that we don't really need anymore.
And they're making things more complicated for some pending work of mine.
Differential Revision: https://phabricator.services.mozilla.com/D89687
The files are copied verbatim from upstream autoconf 2.13 (but only the
files we need) and old.configure is adapted to use the vendored version.
Differential Revision: https://phabricator.services.mozilla.com/D89554
When configuring without system NSPR the configuration variable
PKGCONF_REQUIRES_PRIVATE isn't ever set, leading to a .pc file that still
contains the @PKGCONF_REQUIRES_PRIVATE@ stub.
Ensure that we always define PKGCONF_REQUIRES_PRIVATE, by setting it to an
empty string in case no system-nsrp is enabled.
In this way, the pkg-config file stub will be always replaced.
Differential Revision: https://phabricator.services.mozilla.com/D88179
For not-well-understood reasons, ld's `--gc-sections` discards a large number of the PGO bookkeeping structures that enable us to keep track of function counters, and the effect gets worse in object files generated by clang-10.
As much as I'd like to understand this better, the investigations take way too much time. As a path of least resistance, we can disable `--gc-sections` for the instrumentation phase of PGO builds. It won't harm anything since users never see those builds, and it will improve the performance of the optimized phase greatly.
Differential Revision: https://phabricator.services.mozilla.com/D78112
We don't anticipate end users will actually care to do this, but it's useful especially for unit tests. For example, after bug 1659539, Python `configure` tests will run in a new, non-`init_py3` `virtualenv`, and we'll want to target that `virtualenv` for `configure` rather than having it create a new `virtualenv` for no reason.
Differential Revision: https://phabricator.services.mozilla.com/D88661
2020-08-21 Kevin Jacobs <kjacobs@mozilla.com>
* automation/abi-check/previous-nss-release, lib/nss/nss.h,
lib/softoken/softkver.h, lib/util/nssutil.h:
Set version numbers to 3.57 Beta
[783f49ae6126]
2020-08-24 Kevin Jacobs <kjacobs@mozilla.com>
* gtests/ssl_gtest/ssl_auth_unittest.cc, lib/ssl/dtls13con.c,
lib/ssl/dtlscon.c, lib/ssl/ssl3con.c, lib/ssl/sslimpl.h,
lib/ssl/sslnonce.c:
Bug 1653641 - Cleanup inaccurate DTLS comments, code review fixes.
r=mt
[0e1b5c711cb9]
2020-08-24 Robert Relyea <rrelyea@redhat.com>
* lib/freebl/fipsfreebl.c, lib/softoken/fipstest.c,
lib/softoken/kbkdf.c, lib/softoken/lowpbe.c, lib/softoken/lowpbe.h,
lib/softoken/pkcs11c.c, lib/softoken/pkcs11i.h,
lib/softoken/sftkhmac.c, lib/softoken/sftkike.c:
Bug 1660304 New FIPS IG requires self-tests for approved kdfs.
r=ueno comments=kjacobs
FIPS guidance now requires self-tests for our kdfs. It also requires
self-tests for cmac which we didn't have in the cmac patch.
Currently only one test per kdf is necessary. Specifially for
SP-800-108, only one of the three flavors are needed (counter,
feedback, or pipeline). This patch includes more complete testing
but it has been turned off the currently extraneous tests under the
assumption that NIST guidance may require them in the future. HKDF
is currently not included in FIPS, but is on track to be included,
so hkdf have been included in this patch.
Because the test vectors are const strings, the patch pushes some
const definitions that were missing in existing private interfaces.
There are three flavors of self-tests: Function implemented in
freebl are added to the freebl/fipsfreebl.c Functions implemented in
pkcs11c.c have selftests completely implemented in
softoken/fipstest.c Functions implemented in their own .c file have
their selftest function implemented in that .c file and called by
fipstests.c These are consistant with the previous choices for
selftests.
Some private interfaces that took in keys from pkcs #11 structures
or outputted keys to pkcs #11 structures were modified to optionally
take keys in by bytes and output keys as bytes so the self-tests can
work in just bytes.
[5dca54fe61c2]
2020-08-25 Daiki Ueno <dueno@redhat.com>
* lib/softoken/manifest.mn:
Bug 1659252, disable building libnssdbm3.so if NSS_DISABLE_DBM=1,
r=rrelyea
Reviewers: rrelyea
Reviewed By: rrelyea
Bug #: 1659252
[4d55d36ca6ef]
2020-08-24 Kevin Jacobs <kjacobs@mozilla.com>
* lib/pk11wrap/pk11cxt.c, lib/softoken/pkcs11c.c, lib/softoken/sdb.c,
lib/softoken/sftkpwd.c:
Bug 1651834 - Fix various static analyzer warnings. r=rrelyea
[ab04fd73fd6d]
2020-08-28 Mike Hommey <mh@glandium.org>
* lib/freebl/blapii.h:
Bug 1661810 - Define pre_align/post_align based on the compiler.
r=jcj
Things worked fine before we upgraded to clang 11 presumably because
the stack was always 16-bytes aligned in the first place, or
something akin to that, and the lack of pre_align/post_align doing
anything didn't matter. The runtime misalignment of the stack may
well be a clang > 9 bug, but keeping pre_align/post_align tied to
the x86/x64 is a footgun anyways.
[c100e11991f6] [tip]
Differential Revision: https://phabricator.services.mozilla.com/D88876
config.guess infers information about the compiler using environment
variables, such as CC. However, we use such environment variables to
configure the tooling for the target.
Differential Revision: https://phabricator.services.mozilla.com/D88085
Previously, we would optimistically attempt to use a Rust toolchain that
matches the current C toolchain, and would throw an error if an
attempted compile with that Rust toolchain failed.
Instead, if we fail to detect a usable Rust toolchain, we now helpfully
inform users of their two options: change C toolchain, or install
matching Rust toolchain.
Differential Revision: https://phabricator.services.mozilla.com/D88084
config.guess infers information about the compiler using environment
variables, such as CC. However, we use such environment variables to
configure the tooling for the target.
Differential Revision: https://phabricator.services.mozilla.com/D88085
Previously, we would optimistically attempt to use a Rust toolchain that
matches the current C toolchain, and would throw an error if an
attempted compile with that Rust toolchain failed.
Instead, if we fail to detect a usable Rust toolchain, we now helpfully
inform users of their two options: change C toolchain, or install
matching Rust toolchain.
Differential Revision: https://phabricator.services.mozilla.com/D88084
For not-well-understood reasons, ld's `--gc-sections` discards a large number of the PGO bookkeeping structures that enable us to keep track of function counters, and the effect gets worse in object files generated by clang-10.
As much as I'd like to understand this better, the investigations take way too much time. As a path of least resistance, we can disable `--gc-sections` for the instrumentation phase of PGO builds. It won't harm anything since users never see those builds, and it will improve the performance of the optimized phase greatly.
Differential Revision: https://phabricator.services.mozilla.com/D78112
For not-well-understood reasons, ld's `--gc-sections` discards a large number of the PGO bookkeeping structures that enable us to keep track of function counters, and the effect gets worse in object files generated by clang-10.
As much as I'd like to understand this better, the investigations take way too much time. As a path of least resistance, we can disable `--gc-sections` for the instrumentation phase of PGO builds. It won't harm anything since users never see those builds, and it will improve the performance of the optimized phase greatly.
Differential Revision: https://phabricator.services.mozilla.com/D78112
We have a minimum rust version required for compilations. For both stable and beta rust compilers, we can trust that they will have all the stabilized features we're expecting.
However, for nightlies, they may "match" our minimum version, but may have been released in the version window before a certain feature we need has been stabilized.
So, when validating rustc version in configure, ensure that the nightly is at least one version newer than our expected version.
Differential Revision: https://phabricator.services.mozilla.com/D86889
In two different places we've been encountering issues regarding 1) how we configure the system Python environment and 2) how the system Python environment relates to the `virtualenv`s that we use for building, testing, and other dev tasks. Specifically:
1. With the push to use `glean` for telemetry in `mach`, we are requiring (or rather, strongly encouraging) the `glean_sdk` Python package to be installed with bug 1651424. `mach bootstrap` upgrades the library using your system Python 3 in bug 1654607. We can't vendor it due to the package containing native code. Since we generally vendor all code required for `mach` to function, requiring that the system Python be configured with a certain version of `glean` is an unfortunate change.
2. The build uses the vendored `glean_parser` for a number of build tasks. Since the vendored `glean_parser` conflicts with the globally-installed `glean_sdk` package, we had to add special ad-hoc handling to allow us to circumvent this conflict in bug 1655781.
3. We begin to rely more and more on the `zstandard` package during build tasks, this package again being one that we can't vendor due to containing native code. Bug 1654994 contained more ad-hoc code which subprocesses out from the build system's `virtualenv` to the SYSTEM `python3` binary, assuming that the system `python3` has `zstandard` installed.
As we rely more on `glean_sdk`, `zstandard`, and other packages that are not vendorable, we need to settle on a standard model for how `mach`, the build process, and other `mach` commands that may make their own `virtualenv`s work in the presence of unvendorable packages.
With that in mind, this patch does all the following:
1. Separate out the `mach` `virtualenv_packages` from the in-build `virtualenv_packages`. Refactor the common stuff into `common_virtualenv_packages.txt`. Add functionality to the `virtualenv_packages` manifest parsing to allow the build `virtualenv` to "inherit" from the parent by pointing to the parent's `site-packages`. The `in-virtualenv` feature from bug 1655781 is no longer necessary, so delete it.
2. Add code to `bootstrap`, as well as a new `mach` command `create-mach-environment` to create `virtualenv`s in `~/.mozbuild`.
3. Add code to `mach` to dispatch either to the in-`~/.mozbuild` `virtualenv`s (or to the system Python 3 for commands which cannot run in the `virtualenv`s, namely `bootstrap` and `create-mach-environment`).
4. Remove the "add global argument" feature from `mach`. It isn't used and conflicts with (3).
5. Remove the `--print-command` feature from `mach` which is obsoleted by these changes.
This has the effect of allowing us to install packages that cannot be vendored into a "common" place (namely the global `~/.mozbuild` `virtualenv`s) and use those from the build without requiring us to hit the network. Miscellaneous implementation notes:
1. We allow users to force running `mach` with the system Python if they like. For now it doesn't make any sense to require 100% of people to create these `virtualenv`s when they're allowed to continue on with the old behavior if they like. We also skip this in CI.
2. We needed to duplicate the global-argument logic into the `mach` script to allow for the dispatch behavior. This is something we avoided with the Python 2 -> Python 3 migration with the `--print-command` feature, justifying its use by saying it was only temporarily required until all `mach` commands were running with Python 3. With this change, we'll need to be able to determine the `mach` command from the shell script for the forseeable future, and committing to this forever with the cost that `--print-command` incurs (namely `mach` startup time, an additional .4s on my machine) didn't seem worth it to me. It's not a ton of duplicated code.
Differential Revision: https://phabricator.services.mozilla.com/D85916
This makes us use one less tool from MSVC, and removes one more use of wine
in cross builds.
We replace the call to either rc/llvm-rc or windres with a wrapper script.
While the script is not strictly needed for the latter, we use a wrapper
in that case anyway because it's one step towards fixing bug 1498414.
For llvm-rc, however, we need a wrapper because llvm-rc doesn't preprocess
on its own, so the wrapper does that too.
The wrapper script also allows to deal with the default flags passed to
llvm-rc or windres, rather than inherit them from old-configure.
We also need to explicitly pass the codepage to llvm-rc, which was not
necessary with rc (presumably, llvm-rc has a different default).
While here, remove the unused WINDRES subst from js/src/old-configure.in.
Also, while here, we remove --use-temp-file, because as described in the
linked bug and in the windres manual page, it was used to work around bugs
on Windows 98 and earlier.
Differential Revision: https://phabricator.services.mozilla.com/D86312
I noticed that the `objdir:build` entry in `build/virtualenv_packages.txt` entry was apparently unused. This originates from bug 841713, seven years ago, when the `objdir` handling was introduced. Today, this doesn't appear to be serving a purpose. There is no Python library in my `$objdir/build` directory; nor can I see anything in `build/moz.build` or any related files suggesting one could ever appear. I can only assume this feature has outlived its usefulness, so delete it and the relevant in-tree support.
This necessitates slightly changing the signature and implementation of the `activate_pipenv()` method; also update all callers.
Differential Revision: https://phabricator.services.mozilla.com/D85635
I noticed that the `objdir:build` entry in `build/virtualenv_packages.txt` entry was apparently unused. This originates from bug 841713, seven years ago, when the `objdir` handling was introduced. Today, this doesn't appear to be serving a purpose. There is no Python library in my `$objdir/build` directory; nor can I see anything in `build/moz.build` or any related files suggesting one could ever appear. I can only assume this feature has outlived its usefulness, so delete it and the relevant in-tree support.
This necessitates slightly changing the signature and implementation of the `activate_pipenv()` method; also update all callers.
Differential Revision: https://phabricator.services.mozilla.com/D85635
We are currently relying on two different ways to insert side-by-side
manifests in binaries on Windows: through resource files, or through
the use of MT. The latter is not supported on mingw builds, which is
not great.
Link.exe has options to add a manifest at link time without relying on
either method above, but that's not supported on mingw either.
So the best we can do is to move everything to using resource files.
This also avoids using MT, which, on cross builds, requires using wine.
Ideally, the manifests would be declared in moz.build, but that
complicates things for cases like TestDllInterceptor, where there are
multiple binaries in the same directory, but only one of them needs the
manifest. This keeps the status quo of getting the manifest
automatically from the source directory.
Differential Revision: https://phabricator.services.mozilla.com/D85382
Noemi Erli
Agi Sferro
Dorel Luca
Mike Hommey
Aaron Klotz
Mitchell Hentges
David Major
Kevin Jacobs
Rob Wu
Connor Sheehan
Sylvestre Ledru
Bogdan Tara
Emilio Cobos Álvarez
Sebastian Hengst
Dave Townsend
Christian Holler
Cosmin Sabou
Butkovits Atila
Steve Fink
Ricky Stewart
Jed Davis
J.C. Jones
Razvan Maries
Kartikaya Gupta
Marco Trevisan
Narcis Beleuzu
Hector Zhao
Mihai Alexandru Michis