Expand cudacompat hook to always inspect libcuda.so ELF header if available#1784
Merged
Merged
Conversation
Coverage Report for CI Build 24748865496Coverage increased (+0.2%) to 43.481%Details
Uncovered Changes
Coverage RegressionsNo coverage regressions found. Coverage Stats
💛 - Coveralls |
elezar
reviewed
Apr 20, 2026
elezar
reviewed
Apr 20, 2026
cdesiniotis
force-pushed
the
fwd-compat-elf-header
branch
3 times, most recently
from
d3cd1df to
ee08e9a
Compare
cdesiniotis
force-pushed
the
fwd-compat-elf-header
branch
2 times, most recently
from
0a99c5a to
d8b823d
Compare
tariq1890
reviewed
Jun 25, 2026
tariq1890
reviewed
Jul 3, 2026
…ilable This commit reverts NVIDIA@08bf583 which made it so that the libcuda.so ELF header was only ever inspected on certain Tegra systems, like Orin. We now leverage the libcuda.so ELF header in all cases (if available) to determine if the CUDA compat libraries bundled in the container should be used in favor of the host driver libraries. This allows us to support CUDA minor version compatibility. Signed-off-by: Christopher Desiniotis <cdesiniotis@nvidia.com>
…n version strings
Any version string passed as an argument to semver.Compare() must be a valid semantic
version. It is common for NVIDIA driver versions to have leading zeros in the MINOR or
PATCH portion of a version string, e.g. 575.57.08. As a result, a call to
semver.Compare("575.57.08", "575.10.10") would incorrectly return -1 because the first
argument is not a valid semantic version.
And from https://pkg.go.dev/golang.org/x/mod/semver#Compare:
'''
An invalid semantic version string is considered less than a valid one.
All invalid semantic version strings compare equal to each other.
'''
Signed-off-by: Christopher Desiniotis <cdesiniotis@nvidia.com>
cdesiniotis
force-pushed
the
fwd-compat-elf-header
branch
2 times, most recently
from
0a695d6 to
4d240d2
Compare
tariq1890
reviewed
Jul 4, 2026
Contributor
|
Thanks @cdesiniotis, added one more comment. |
Signed-off-by: Christopher Desiniotis <cdesiniotis@nvidia.com>
Signed-off-by: Christopher Desiniotis <cdesiniotis@nvidia.com>
cdesiniotis
force-pushed
the
fwd-compat-elf-header
branch
from
e129775 to
e20df13
Compare
tariq1890
approved these changes
Jul 6, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This commit reverts 08bf583 which made it so that the libcuda.so ELF header was only ever inspected on certain Tegra systems, like Orin. We now leverage the libcuda.so ELF header in all cases (if available) to determine if the CUDA compat libraries bundled in the container should be used in favor of the host driver libraries. This allows us to support CUDA minor version compatibility.