UniqAI/nvidia-container-toolkit
1
0
Fork 0
mirror of https://github.com/NVIDIA/nvidia-container-toolkit synced 2024-12-03 09:34:48 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix bug when using just-in-time CDI spec generation

Browse Source 
This change fixes a bug when using just-in-time CDI spec generation for the
NVIDIA Container Runtime for specific devices (i.e. not 'all').
Instead of unconditionally using the default nvsandboxutils library -- leading
to errors due to undefined symbols -- we check whether the library can be
properly initialised before continuing.

Signed-off-by: Evan Lezar <elezar@nvidia.com>
This commit is contained in:
Evan Lezar 2024-10-30 11:50:16 +01:00
parent d78868cd31
commit 75376d3df2
No known key found for this signature in database
1 changed files with 15 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
pkg/nvcdi/lib-nvml.go
View File

@ -103,7 +103,7 @@ func (l *nvmllib) GetDeviceSpecsByID(ids ...string) ([]specs.Device, error) {
return l.GetDeviceSpecsBy(identifiers...)
}
// GetDeviceSpecsBy is not supported for the gdslib specs.
// GetDeviceSpecsBy returns the device specs for devices with the specified identifiers.
func (l *nvmllib) GetDeviceSpecsBy(identifiers ...device.Identifier) ([]specs.Device, error) {
for _, id := range identifiers {
if id == "all" {
@ -118,10 +118,23 @@ func (l *nvmllib) GetDeviceSpecsBy(identifiers ...device.Identifier) ([]specs.De
}
defer func() {
if r := l.nvmllib.Shutdown(); r != nvml.SUCCESS {
l.logger.Warningf("failed to shutdown NVML: %w", r)
l.logger.Warningf("failed to shutdown NVML: %v", r)
}
}()
if l.nvsandboxutilslib != nil {
if r := l.nvsandboxutilslib.Init(l.driverRoot); r != nvsandboxutils.SUCCESS {
l.logger.Warningf("Failed to init nvsandboxutils: %v; ignoring", r)
l.nvsandboxutilslib = nil
}
defer func() {
if l.nvsandboxutilslib == nil {
return
}
_ = l.nvsandboxutilslib.Shutdown()
}()
}
nvmlDevices, err := l.getNVMLDevicesByID(identifiers...)
if err != nil {
return nil, fmt.Errorf("failed to get NVML device handles: %w", err)