On Tue, Nov 16, 2021 at 12:26:17PM +0100, Vlastimil Babka wrote:

On 10/29/21 05:05, Rustam Kovhaev wrote:

quoted

Let's prepend both kmalloc() and kmem_cache_alloc() allocations with the
size header.
It simplifies the slab API and guarantees that both kmem_cache_alloc()
and kmalloc() memory could be freed by kfree().

meminfo right after the system boot, x86-64 on xfs, without the patch:
Slab:              35456 kB

the same, with the patch:
Slab:              36100 kB

Link: https://lore.kernel.org/lkml/20210929212347.1139666-1-rkovhaev@gmail.com (local)
Signed-off-by: Rustam Kovhaev <redacted>

Sorry for the late reply. I think we can further simplify this. We have:

static void *slob_alloc(size_t size, gfp_t gfp, int align,
			int node, int align_offset)

Previously there was one caller that passed size as unchanged, align_offset
= 0, the other passed size + SLOB_HDR_SIZE, align_offset = SLOB_HDR_SIZE.
Now both callers do the latter. We can drop the align_offset parameter and
pass size unchanged. slob_alloc() can internally add SLOB_HDR_SIZE to size,
and use SLOB_HDR_SIZE instead of align_offset.

Thank you, I'll send a v4.