Imperial-MEng/pintos_22
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update fd_hash to use the fd itself as the hash value for performance, w/ G & E

Browse Source
This commit is contained in:
sBubshait
2024-11-15 15:53:01 +00:00
parent 6b1dbdd34f
commit ea3b3594ea
1 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
src/userprog/syscall.c
View File

@@ -406,7 +406,12 @@ syscall_close (int fd)
unsigned unsigned
fd_hash (const struct hash_elem *element, void *aux UNUSED) fd_hash (const struct hash_elem *element, void *aux UNUSED)
{ {
return hash_int (hash_entry (element, struct open_file, elem)->fd); /* We use the FD as the hash value. This is because the FD is incremented
sequentially and is therefore unique for each file. It positively affects
the performance of the hash table: 1. It is unique so no need to call
expensive hash functions. 2. It being sequential means that the hash table
is more likely to be weight balanced. */
return hash_entry (element, struct open_file, elem)->fd;
} }
/* Comparator function for the open_file table. Compares two entries based on /* Comparator function for the open_file table. Compares two entries based on