Mimalloc vs dlmalloc #20936
Unanswered
technopedi
asked this question in
Q&A
Mimalloc vs dlmalloc
#20936
Replies: 3 comments 23 replies
-
|
It could be a bug in mimalloc, or it could be existing memory corruption bug doesn't show up except under certain memory layout conditions. Can you try |
Beta Was this translation helpful? Give feedback.
7 replies
-
|
I get issue in mimalloc with below stack. Can we say that memory was exhausted |
Beta Was this translation helpful? Give feedback.
-
|
An access out of bounds sounds like a bug in mimalloc or our port of it. Can you provide a testcase? The bug could be related to memory being exhausted, if there is a code path that doesn't handle OOM properly. |
Beta Was this translation helpful? Give feedback.
-
|
How can new give memory access out of bounds? Shouldn't it be coming when trying to access something out of buffer allocated? |
Beta Was this translation helpful? Give feedback.
-
|
A bug inside |
Beta Was this translation helpful? Give feedback.
-
|
I'm trying to figure out the best test case and reduce the application into this error.Because unfortunately the asan+mimalloc in my application is giving heap_buffer_overflow in starting on main itself, due to which unable to use asan to figure out memory access out of bounds issue. Meanwhile is it safe to assume that if list of heaps per thread is exhausted, then block->next will ideally return garbage value, resulting in memory access out of bounds? |
Beta Was this translation helpful? Give feedback.
-
|
After switching to |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
In a simple single thread test that allocates memory 10 MByte at a time I see with This could just as well be the documented worst case performance of +25% memory: https://microsoft.github.io/mimalloc/bench.html The emsdk version I used for the test was v4.0.0 With multithreading (23 threads): |
Beta Was this translation helpful? Give feedback.
16 replies
-
It sounds like maybe mimalloc also doesn't know how to shink its arena-increasing request when once fails? i..e when its tries to make a huge allocation to increase the arena size, and it fails, shouldn't its lower the size of the request until it succeeds? That seems like the only way to actually use all of the memory. |
Beta Was this translation helpful? Give feedback.
-
|
@Markus87 Good idea to compare to native 32 bit! I'm curious to see what you find. @sbc100 I didn't look into that fully, but yeah, I agree it seems natural to try smaller allocations. Let's see what @Markus87 finds on native 32 bit. We can open an issue on mimalloc based on that. |
Beta Was this translation helpful? Give feedback.
-
|
I tested it with mimalloc 2.1.9 for Windows x86 32Bit and it looks similar to wasm32. The output with mimalloc verboseThere seem to be a few issues on mimalloc side that would like to limit the maximum allocated amount for different reasons: I experimented a bit with reserving and commiting ahead but I could not improve on the initial result yet. |
Beta Was this translation helpful? Give feedback.
-
|
I opened an issue microsoft/mimalloc#996 |
Beta Was this translation helpful? Give feedback.
-
|
It might be worth re-checking this with PR #23037 and/or mimalloc v3.0-alpha (work-in-progress changeset here: HEAD...kleisauke:mimalloc-update-3.0.1). |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
A multithreaded application is causing memory access out of bounds with Mimalloc , but doesn't cause any issue with any other allocator like dlmalloc. I have enabled pthreads as well. What could be the possible reason of same?
Beta Was this translation helpful? Give feedback.
All reactions