gbowne1 version of implementation with more implementations done #4

gbowne1 · 2024-02-14T00:30:43-08:00

gbowne1 commented

2024-02-14 00:30:43 -08:00

Added a bunch of stuff to show you how I would do some of this. It might need some cleanup work but hope this helps.

Added better fragment handling
Fixed heap extension logic
Added some minor error handling - Needs more work

I aligned for 64 bit systems which I looked might use ALIGN16 in place of ALIGN4 or even 8. I wasnt sure if you wanted to target just 64 bit.

should be reasonably clean at this point.

some stuff to think about Havent had a lot of time in a couple days.

Added a bunch of stuff to show you how I would do some of this. It might need some cleanup work but hope this helps. Added better fragment handling Fixed heap extension logic Added some minor error handling - Needs more work I aligned for 64 bit systems which I looked might use ALIGN16 in place of ALIGN4 or even 8. I wasnt sure if you wanted to target just 64 bit. should be reasonably clean at this point. some stuff to think about Havent had a lot of time in a couple days.

gbowne1 added 1 commit 2024-02-14 00:30:43 -08:00

gbowne1 version of implementation with more implementations done 19b2e31119

Karutoh commented

2024-02-14 15:00:32 -08:00

Ideally our target should be for both x86 and x64. Just for your operating system and embedded hardware. We want this to be a good example for both.

Karutoh commented

2024-02-14 15:05:45 -08:00

Seems an alignment of 16 bytes is indeed required of x64 architectures.

Karutoh commented

2024-02-14 15:10:50 -08:00

I noticed on line number 34 there's an issue. You've done s += BLOCK_SIZE;. This not ideal and should only hold the size of data the memory block contains, not including the memory block header. Otherwise it will require an extra operation or more.

I noticed on line number 34 there's an issue. You've done `s += BLOCK_SIZE;`. This not ideal and should only hold the size of data the memory block contains, not including the memory block header. Otherwise it will require an extra operation or more.

Karutoh commented

2024-02-14 15:19:38 -08:00

There's an additional ALIGN16 macro function call in fragment_block function which is unnecessary.

There's an additional `ALIGN16` macro function call in `fragment_block` function which is unnecessary.

Karutoh commented

2024-02-14 15:25:36 -08:00

In the find_best_fit function on line number 111. There's a null pointer error. In the statement if (!best_fit || current->size < best_fit->size) will still try the current->size < best_fit->size logic regardless, which best_fit can and will be NULL.

In the `find_best_fit` function on line number 111. There's a null pointer error. In the statement `if (!best_fit || current->size < best_fit->size)` will still try the `current->size < best_fit->size` logic regardless, which `best_fit` can and will be `NULL`.

Karutoh commented

2024-02-14 15:39:19 -08:00

In my_custom_free there's an error on line number 232 for the statement if (b == last). The last global variable was not updated prior to that statement, which in turn the statement will never be true and may point to stale data on the heap.

In `my_custom_free` there's an error on line number 232 for the statement `if (b == last)`. The `last` global variable was not updated prior to that statement, which in turn the statement will never be true and may point to stale data on the heap.

Karutoh self-assigned this 2024-02-14 15:41:21 -08:00

Karutoh requested review from Karutoh 2024-02-14 15:41:33 -08:00

Karutoh requested changes 2024-02-14 15:44:23 -08:00

Karutoh left a comment

This is not an error. The changes of struct block* b = (struct block*)ptr - 1; to struct block *b = (struct block *)((char *)ptr - BLOCK_SIZE); adds more text to parse for the compiler and the user. Both do the same thing.

Please resolve the errors provided above.

This is not an error. The changes of `struct block* b = (struct block*)ptr - 1;` to `struct block *b = (struct block *)((char *)ptr - BLOCK_SIZE);` adds more text to parse for the compiler and the user. Both do the same thing. Please resolve the errors provided above.

Karutoh removed their assignment 2024-02-14 15:44:57 -08:00

gbowne1 was assigned by Karutoh

2024-02-14 15:45:00 -08:00

gbowne1 added 1 commit 2024-02-14 16:23:22 -08:00

fixing issues from PR fe9d9e3836

gbowne1 added 1 commit 2024-02-14 16:51:52 -08:00

Merge remote-tracking branch 'origin/main' into gbowne1_impl f83a00f4d5

gbowne1 commented

2024-02-14 16:52:21 -08:00

yey me
fixed it.

yey me fixed it.

gbowne1 added 1 commit 2024-02-14 16:55:43 -08:00