-
-
Notifications
You must be signed in to change notification settings - Fork 14.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update request: python310Packages.tensorflow 2.10.1 → 2.12.0 #224353
Comments
FWIW I would welcome not being listed as maintainer anymore. |
I have some bazel expertise and I've been helping @samuela with jax related packages now and then, would be happy to help out or give advice around bazel specifics, but unfortunately don't know much about tensorflow. We also have a |
Started doing the update and got the hashes for all systems. No luck with system libraries (absl-py, protobuf) and it generally fails during the build. Can't sink more time into this. commit 4837b55374d331894431fc0d4088bae3bd222e75
Author: Martin Weinelt <hexa@darmstadt.ccc.de>
Date: Tue Apr 4 17:07:09 2023 +0200
WIP: tensorflow 2.12.0
diff --git a/pkgs/development/python-modules/tensorflow/default.nix b/pkgs/development/python-modules/tensorflow/default.nix
index d2549f7b057f..3efc9ca68b4e 100644
--- a/pkgs/development/python-modules/tensorflow/default.nix
+++ b/pkgs/development/python-modules/tensorflow/default.nix
@@ -9,7 +9,7 @@
, termcolor, grpcio, six, wrapt, protobuf-python, tensorflow-estimator-bin
, dill, flatbuffers-python, portpicker, tblib, typing-extensions
# Common deps
-, git, pybind11, which, binutils, glibcLocales, cython, perl, coreutils
+, git, pybind11, which, binutils, glibcLocales, cython, perl
# Common libraries
, jemalloc, mpi, gast, grpc, sqlite, boringssl, jsoncpp, nsync
, curl, snappy, flatbuffers-core, lmdb-core, icu, double-conversion, libpng, libjpeg_turbo, giflib, protobuf-core
@@ -98,7 +98,7 @@ let
tfFeature = x: if x then "1" else "0";
- version = "2.11.0";
+ version = "2.12.0";
variant = lib.optionalString cudaSupport "-gpu";
pname = "tensorflow${variant}";
@@ -207,7 +207,7 @@ let
owner = "tensorflow";
repo = "tensorflow";
rev = "refs/tags/v${version}";
- hash = "sha256-OYh61/83yv+ycivylfdS8yFUIUAk8euAPvmfjPzldGs=";
+ hash = "sha256-8e4skLGKUtJ75ldE58jVLg9pusZKrdTMSORpb3C1MxQ=";
};
# On update, it can be useful to steal the changes from gentoo
@@ -260,16 +260,19 @@ let
# list of valid syslibs in
# https://github.com/tensorflow/tensorflow/blob/master/third_party/systemlibs/syslibs_configure.bzl
TF_SYSTEM_LIBS = lib.concatStringsSep "," ([
- "absl_py"
+ # ERROR: /build/source/tensorflow/python/tools/BUILD:322:11: no such target '@absl_py//absl/flags:argparse_flags': target 'argparse_flags' not declared in package 'absl/flags' defined by /build/output/external/absl_py/absl/flags/BUILD and referenced by '//tensorflow/python/tools:saved_model_cli_lib'
+ # "absl_py"
"astor_archive"
"astunparse_archive"
"boringssl"
# Not packaged in nixpkgs
- # "com_github_googleapis_googleapis"
# "com_github_googlecloudplatform_google_cloud_cpp"
"com_github_grpc_grpc"
- "com_google_protobuf"
- # Fails with the error: external/org_tensorflow/tensorflow/core/profiler/utils/tf_op_utils.cc:46:49: error: no matching function for call to 're2::RE2::FullMatch(absl::lts_2020_02_25::string_view&, re2::RE2&)'
+ # ERROR: /build/source/tensorflow/compiler/xla/stream_executor/BUILD:466:11: no such target '@com_google_absl//absl/functional:any_invocable': target 'any_invocable' not declared in package 'absl/functional' defined by /build/output/external/com_google_absl/absl/functional/BUILD.bazel and referenced by '//tensorflow/compiler/xla/stream_executor:kernel'
+ # "com_google_absl"
+ # ERROR: /build/source/tensorflow/core/protobuf/tpu/BUILD:13:17: no such target '@com_google_protobuf//:well_known_types_py_pb2_genproto': target 'well_known_types_py_pb2_genproto' not declared in package '' defined by /build/output/external/com_google_protobuf/BUILD.bazel and referenced by '//tensorflow/core/protobuf/tpu:tpu_embedding_configuration_proto_py_genproto'
+ # "com_google_protobuf"
+ # ERROR: /build/source/tensorflow/core/data/service/BUILD:885:11: Compiling tensorflow/core/data/service/url.cc failed: (Exit 1): gcc failed: error executing command /nix/store/h5003wsy3qqimqvrkn3bc5mwq4hhidag-gcc-wrapper-12.2.0/bin/gcc -U_FORTIFY_SOURCE -fstack-protector -Wall -Wunused-but-set-parameter -Wno-free-nonheap-object -fno-omit-frame-pointer -g0 -O2 ... (remaining 45 arguments skipped)
# "com_googlesource_code_re2"
"curl"
"cython"
@@ -394,11 +397,11 @@ let
fetchAttrs = {
sha256 = {
x86_64-linux = if cudaSupport
- then "sha256-/wB9EpaDPg3TrD9qggdA4vPgzvmaKc6dDnLjoYTJC5o="
- else "sha256-QgOaUaq0V5HG9BOv9nEw8OTSlzINNFvbnyP8Vx+r9Xw=";
- aarch64-linux = "sha256-zjnRtTG1j9cZTbP0Xnk2o/zWTNsP8T0n4Ai8IiAT3PE=";
- x86_64-darwin = "sha256-RBLox9rzBKcZMm4NwnT7vQ/EjapWQJkqxuQ0LIdaM1E=";
- aarch64-darwin = "sha256-BRzh79lYvMHsUMk8BEYDLHTpnmeZ9+0lrDtj4XI1YY4=";
+ then "sha256-GIlHDvHZOQO2z98qv+T+MkLXEqHoWcou3kz1gjagGvE="
+ else "sha256-H5hnl5hs9LCiUNFzQKekUsf25HxcgJ6VmM09YOlZoBI=";
+ aarch64-linux = "sha256-5OVaflKWSP9ahoAYzgW5H2wiZzj40m+Blb/4L5/F1yw=";
+ x86_64-darwin = "sha256-sWBpNm60zVLlpV4qIeQGA02x4plp7xPaOAqWO5B55JQ=";
+ aarch64-darwin = "sha256-lPC75KJhKClFdBagp4y/PdbHU8hB/Z+NzNoFnr2Gtm4=";
}.${stdenv.hostPlatform.system} or (throw "unsupported system ${stdenv.hostPlatform.system}");
}; |
The day has come: TF has been marked as insecure and is not building (https://github.com/samuela/nixpkgs-upkeep/actions/runs/4613477016/jobs/8155492157) |
This issue has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/tensorflow-unstable-not-installing-after-update/27041/2 |
Looks like this can be closed: #255998 |
Notify maintainers
Look, I know it's total Bazel, but we need to update our tensorflow; @mweinelt also points out there are 19CVEs that would be addressed by the update (tf marked insecure in #224436). I have not managed to allocate time for this myself yet, so I'm making a tracking issue for now instead. I'm also pinging our potential heroes:
CC @NixOS/cuda-maintainers because we don't want to be left without tensorflow
Declared package maintainers:
CC @jypCC @abbradar
Recent committers from the git blame:
CC @veprbl
CC @symphorien
CC @Stunkymonkey
CC @tricktron
CC @Artturin
Note for maintainers: Please tag this issue in your PR.
Updates
EDIT 2023-04-06: #224846 can be merged faster and should address the CVEs
EDIT 2023-04-21: #224846 has been merged and backported, so CVEs are not an issue anymore. Kudos to @ risicle!
The text was updated successfully, but these errors were encountered: