Use semantic versioning #3
Merged
Conversation
This file contains 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
Closed
Closed
minsii
pushed a commit
to minsii/nccl
that referenced
this pull request
Sep 16, 2022
Co-authored-by: Facebook Community Bot <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 9, 2023
ncclUniqueId is an opaque array of bytes that is casted by
ncclCommInitRankFunc() to struct ncclBootstrapHandle. Which is required to be
aligned to 8 bytes in memory for defined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
bootstrap.cc:242:40: runtime error: member access within misaligned address 0x61100043846c for type 'struct ncclBootstrapHandle', which requires 8 byte alignment
0x61100043846c: note: pointer points here
5f 00 00 00 ce 7d 2b 40 b1 c5 ce 59 02 00 9f 21 0a c1 35 7a 00 00 00 00 00 00 00 00 00 00 00 00
^
#0 0x7feba7d327b4 in bootstrapInit(ncclBootstrapHandle*, ncclComm*) /home/ubuntu/nccl/src/bootstrap.cc:242
NVIDIA#1 0x7feba7c7bd48 in ncclCommInitRankFunc /home/ubuntu/nccl/src/init.cc:1350
NVIDIA#2 0x7feba7e0cebc in ncclAsyncJobMain(void*) /home/ubuntu/nccl/src/group.cc:63
NVIDIA#3 0x7feba72e2608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#4 0x7feba6ed4132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 9, 2023
ncclRealloc() is used in the code also to perform the initial allocation.
This results in calling memcpy() with a NULL source pointer and 0 copy size.
Allegedly this seems harmless, but memcpy is declared to never accept NULL pointers.
Which could lead to undefined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
include/alloc.h:68:9: runtime error: null pointer passed as argument 2, which is declared to never be null
#0 0x7f867cf80932 in ncclResult_t ncclRealloc<ncclProxyConnection*>(ncclProxyConnection***, unsigned long, unsigned long) include/alloc.h:68
NVIDIA#1 0x7f867cf6c838 in ncclProxyNewConnection /home/ubuntu/nccl/src/proxy.cc:942
NVIDIA#2 0x7f867cf74ba8 in proxyConnInit /home/ubuntu/nccl/src/proxy.cc:1261
NVIDIA#3 0x7f867cf78021 in proxyProgressAsync /home/liran/nccl/src/proxy.cc:1318
NVIDIA#4 0x7f867cf79ff0 in proxyServiceInitOp /home/liran/nccl/src/proxy.cc:1377
NVIDIA#5 0x7f867cf7c052 in ncclProxyService(void*) /home/ubuntu/nccl/src/proxy.cc:1507
NVIDIA#6 0x7f867c3c8608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#7 0x7f867bfba132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 9, 2023
ncclUniqueId is an opaque array of bytes that is casted by
ncclCommInitRankFunc() to struct ncclBootstrapHandle. Which is required to be
aligned to 8 bytes in memory for defined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
bootstrap.cc:242:40: runtime error: member access within misaligned address 0x61100043846c for type 'struct ncclBootstrapHandle', which requires 8 byte alignment
0x61100043846c: note: pointer points here
5f 00 00 00 ce 7d 2b 40 b1 c5 ce 59 02 00 9f 21 0a c1 35 7a 00 00 00 00 00 00 00 00 00 00 00 00
^
#0 0x7feba7d327b4 in bootstrapInit(ncclBootstrapHandle*, ncclComm*) /home/ubuntu/nccl/src/bootstrap.cc:242
NVIDIA#1 0x7feba7c7bd48 in ncclCommInitRankFunc /home/ubuntu/nccl/src/init.cc:1350
NVIDIA#2 0x7feba7e0cebc in ncclAsyncJobMain(void*) /home/ubuntu/nccl/src/group.cc:63
NVIDIA#3 0x7feba72e2608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#4 0x7feba6ed4132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 9, 2023
ncclRealloc() is used in the code also to perform the initial allocation.
This results in calling memcpy() with a NULL source pointer and 0 copy size.
Allegedly this seems harmless, but memcpy is declared to never accept NULL pointers.
Which could lead to undefined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
include/alloc.h:68:9: runtime error: null pointer passed as argument 2, which is declared to never be null
#0 0x7f867cf80932 in ncclResult_t ncclRealloc<ncclProxyConnection*>(ncclProxyConnection***, unsigned long, unsigned long) include/alloc.h:68
NVIDIA#1 0x7f867cf6c838 in ncclProxyNewConnection /home/ubuntu/nccl/src/proxy.cc:942
NVIDIA#2 0x7f867cf74ba8 in proxyConnInit /home/ubuntu/nccl/src/proxy.cc:1261
NVIDIA#3 0x7f867cf78021 in proxyProgressAsync /home/liran/nccl/src/proxy.cc:1318
NVIDIA#4 0x7f867cf79ff0 in proxyServiceInitOp /home/liran/nccl/src/proxy.cc:1377
NVIDIA#5 0x7f867cf7c052 in ncclProxyService(void*) /home/ubuntu/nccl/src/proxy.cc:1507
NVIDIA#6 0x7f867c3c8608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#7 0x7f867bfba132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 10, 2023
ncclUniqueId is an opaque array of bytes that is casted by
ncclCommInitRankFunc() to struct ncclBootstrapHandle. Which is required
to be aligned to 8 bytes in memory for defined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
bootstrap.cc:242:40: runtime error: member access within misaligned address 0x61100043846c for type 'struct ncclBootstrapHandle', which requires 8 byte alignment
0x61100043846c: note: pointer points here
5f 00 00 00 ce 7d 2b 40 b1 c5 ce 59 02 00 9f 21 0a c1 35 7a 00 00 00 00 00 00 00 00 00 00 00 00
^
#0 0x7feba7d327b4 in bootstrapInit(ncclBootstrapHandle*, ncclComm*) /home/ubuntu/nccl/src/bootstrap.cc:242
NVIDIA#1 0x7feba7c7bd48 in ncclCommInitRankFunc /home/ubuntu/nccl/src/init.cc:1350
NVIDIA#2 0x7feba7e0cebc in ncclAsyncJobMain(void*) /home/ubuntu/nccl/src/group.cc:63
NVIDIA#3 0x7feba72e2608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#4 0x7feba6ed4132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 10, 2023
ncclRealloc() is used in the code also to perform the initial
allocation. This results in calling memcpy() with a NULL source pointer
and 0 copy size.
Allegedly this seems harmless, but memcpy is declared to never accept
NULL pointers. Which could lead to undefined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
include/alloc.h:68:9: runtime error: null pointer passed as argument 2, which is declared to never be null
#0 0x7f867cf80932 in ncclResult_t ncclRealloc<ncclProxyConnection*>(ncclProxyConnection***, unsigned long, unsigned long) include/alloc.h:68
NVIDIA#1 0x7f867cf6c838 in ncclProxyNewConnection /home/ubuntu/nccl/src/proxy.cc:942
NVIDIA#2 0x7f867cf74ba8 in proxyConnInit /home/ubuntu/nccl/src/proxy.cc:1261
NVIDIA#3 0x7f867cf78021 in proxyProgressAsync /home/liran/nccl/src/proxy.cc:1318
NVIDIA#4 0x7f867cf79ff0 in proxyServiceInitOp /home/liran/nccl/src/proxy.cc:1377
NVIDIA#5 0x7f867cf7c052 in ncclProxyService(void*) /home/ubuntu/nccl/src/proxy.cc:1507
NVIDIA#6 0x7f867c3c8608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#7 0x7f867bfba132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 10, 2023
ncclUniqueId is an opaque array of bytes that is casted by
ncclCommInitRankFunc() to struct ncclBootstrapHandle. Which is required
to be aligned to 8 bytes in memory for defined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
bootstrap.cc:242:40: runtime error: member access within misaligned address 0x61100043846c for type 'struct ncclBootstrapHandle', which requires 8 byte alignment
0x61100043846c: note: pointer points here
5f 00 00 00 ce 7d 2b 40 b1 c5 ce 59 02 00 9f 21 0a c1 35 7a 00 00 00 00 00 00 00 00 00 00 00 00
^
#0 0x7feba7d327b4 in bootstrapInit(ncclBootstrapHandle*, ncclComm*) /home/ubuntu/nccl/src/bootstrap.cc:242
NVIDIA#1 0x7feba7c7bd48 in ncclCommInitRankFunc /home/ubuntu/nccl/src/init.cc:1350
NVIDIA#2 0x7feba7e0cebc in ncclAsyncJobMain(void*) /home/ubuntu/nccl/src/group.cc:63
NVIDIA#3 0x7feba72e2608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#4 0x7feba6ed4132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
liralon
added a commit
to liralon/nccl
that referenced
this pull request
Aug 10, 2023
ncclRealloc() is used in the code also to perform the initial
allocation. This results in calling memcpy() with a NULL source pointer
and 0 copy size.
Allegedly this seems harmless, but memcpy is declared to never accept
NULL pointers. Which could lead to undefined behaviour.
This issue was caught by running nccl-tests with NCCL built with UBSAN:
```
include/alloc.h:68:9: runtime error: null pointer passed as argument 2, which is declared to never be null
#0 0x7f867cf80932 in ncclResult_t ncclRealloc<ncclProxyConnection*>(ncclProxyConnection***, unsigned long, unsigned long) include/alloc.h:68
NVIDIA#1 0x7f867cf6c838 in ncclProxyNewConnection /home/ubuntu/nccl/src/proxy.cc:942
NVIDIA#2 0x7f867cf74ba8 in proxyConnInit /home/ubuntu/nccl/src/proxy.cc:1261
NVIDIA#3 0x7f867cf78021 in proxyProgressAsync /home/ubuntu/nccl/src/proxy.cc:1318
NVIDIA#4 0x7f867cf79ff0 in proxyServiceInitOp /home/ubuntu/nccl/src/proxy.cc:1377
NVIDIA#5 0x7f867cf7c052 in ncclProxyService(void*) /home/ubuntu/nccl/src/proxy.cc:1507
NVIDIA#6 0x7f867c3c8608 in start_thread /build/glibc-2.31/nptl/pthread_create.c:477
NVIDIA#7 0x7f867bfba132 in __clone (/lib/x86_64-linux-gnu/libc.so.6)
```
Signed-off-by: Liran Alon <[email protected]>
alexander-zinoviev
pushed a commit
to alexander-zinoviev/nccl
that referenced
this pull request
Nov 7, 2024
[v2.21.5-1] Add our patches to NCCL v2.21.5-1
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Adherence to Semantic Versioning.
Necessary to build a proper deb package (#2).
/cc @borisfom