Skip to content

fix: prevent unsigned integer underflow in NVML memory info hook #201

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.

Sign up for GitHub

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

Open
miaobyte wants to merge 4 commits into
base: main
Choose a base branch
from
+28 −11
Open

fix: prevent unsigned integer underflow in NVML memory info hook #201

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
df81330
fix: prevent unsigned integer underflow when usage exceeds memory limit
Jun 4, 2026
d6f4da8
fix: nvml memory underflow — usage>limit 时 clamp 而非报错
Jul 24, 2026
fee12be
fix: clamp NVML total against physical GPU memory with min(limit, phy…
miaobyte Jul 28, 2026
4c6680b
fix: use actual_limit for NVML clamp and warning
miaobyte Jul 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 5 additions & 4 deletions src/cuda/memory.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -512,16 +512,17 @@
*free = *total - usage;
LOG_INFO("after free=%ld total=%ld", *free, *total);
return CUDA_SUCCESS;
} else if (limit < usage) {
LOG_WARN("limit < usage; usage=%ld, limit=%ld", usage, limit);
return CUDA_ERROR_INVALID_VALUE;
} else {
CUDA_OVERRIDE_CALL(cuda_library_entry,cuMemGetInfo_v2, free, total);
LOG_INFO("orig free=%ld total=%ld limit=%ld usage=%ld",
*free, *total, limit, usage);
// Ensure total memory does not exceed the physical or imposed limit.
size_t actual_limit = (limit > *total) ? *total : limit;
*free = (actual_limit > usage) ? (actual_limit - usage) : 0;
size_t clamped = (usage > limit) ? limit : usage;

Check failure on line 521 in src/cuda/memory.c

View workflow job for this annotation

GitHub Actions / cpplint 

[cpplint] reported by reviewdog 🐶
 Missing username in TODO; it should look like "// TODO(my_username): Stuff."  [readability/todo] [2]

Raw Output:
src/cuda/memory.c:521:  Missing username in TODO; it should look like "// TODO(my_username): Stuff."  [readability/todo] [2]
if (usage > limit) {
LOG_WARN("CUDA meminfo: usage %lu exceeds limit %lu, clamping", usage, limit);

Check failure on line 523 in src/cuda/memory.c

View workflow job for this annotation

GitHub Actions / cpplint 

[cpplint] reported by reviewdog 🐶
 Missing space after ,  [whitespace/comma] [3]

Raw Output:
src/cuda/memory.c:523:  Missing space after ,  [whitespace/comma] [3]

Check failure on line 523 in src/cuda/memory.c

View workflow job for this annotation

GitHub Actions / cpplint 

[cpplint] reported by reviewdog 🐶
 Lines should be <= 120 characters long  [whitespace/line_length] [2]

Raw Output:
src/cuda/memory.c:523:  Lines should be <= 120 characters long  [whitespace/line_length] [2]
}
*free = (actual_limit > clamped) ? (actual_limit - clamped) : 0;
*total = actual_limit;
LOG_INFO("after free=%ld total=%ld limit=%ld usage=%ld",
*free, *total, limit, usage);
Expand Down
30 changes: 23 additions & 7 deletions src/nvml/hook.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -353,18 +353,34 @@ nvmlReturn_t _nvmlDeviceGetMemoryInfo(nvmlDevice_t device,void* memory,int versi
return NVML_SUCCESS;
}
} else {
size_t physical_total;
switch (version) {
case 1:
((nvmlMemory_t*)memory)->free = (limit-usage);
((nvmlMemory_t*)memory)->total = limit;
((nvmlMemory_t*)memory)->used = usage;
physical_total = ((nvmlMemory_t*)memory)->total;
break;
case 2:
physical_total = ((nvmlMemory_v2_t *)memory)->total;
break;
default:
return NVML_ERROR_INVALID_ARGUMENT;
}
size_t actual_limit = (limit > physical_total) ? physical_total : limit;
size_t clamped = (usage > actual_limit) ? actual_limit : usage;
if (usage > actual_limit) {
LOG_WARN("NVML meminfo: usage %lu exceeds limit %lu, clamping", usage, actual_limit);
}
switch (version) {
case 1:
((nvmlMemory_t*)memory)->free = (actual_limit > clamped) ? (actual_limit - clamped) : 0;
((nvmlMemory_t*)memory)->total = actual_limit;
((nvmlMemory_t*)memory)->used = clamped;
return NVML_SUCCESS;
case 2:
((nvmlMemory_v2_t *)memory)->free = (limit-usage);
((nvmlMemory_v2_t *)memory)->total = limit;
((nvmlMemory_v2_t *)memory)->used = usage;
((nvmlMemory_v2_t *)memory)->free = (actual_limit > clamped) ? (actual_limit - clamped) : 0;
((nvmlMemory_v2_t *)memory)->total = actual_limit;
((nvmlMemory_v2_t *)memory)->used = clamped;
return NVML_SUCCESS;
}
}
}
return NVML_SUCCESS;
}
Expand Down
Loading