Thread (43 messages) 43 messages, 4 authors, 2024-01-31

Re: [PATCH 5/6] eventfs: get rid of dentry pointers without refcounts

From: Linus Torvalds <torvalds@linux-foundation.org>
Date: 2024-01-30 23:25:27
Also in: lkml

On Tue, 30 Jan 2024 at 15:10, Steven Rostedt [off-list ref] wrote:
quoted
At which point a name pointer would *just* fit in 96 bytes.
Does that mean I should keep the kstrdup_const()?
You should check that my math matches reality and relevant
configurations, but yes, at 96 bytes that should fit exactly in one
slab entry on both x86-64 and arm64 (now that arm finally does sane
kmalloc sizes - for the longest time they were all 128-byte aligned
due to historical horrible DMA coherency issues).

But you might also add a comment to the eventfs_inode definition,
because if that ever changes, the math changes again. For example,
adding just one more attribute data would make it all fall apart.

If you *really* want to optimize that data structure, I think you can
make the child list be a 'hlist'. That makes the list head ('children'
becomes a 'hlist_head') smaller, even if the list entry ('list'
becomes a 'hlist_node') stays the same size.

             Linus
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help