Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux

mm: vmalloc: implement vrealloc()

Patch series "Align kvrealloc() with krealloc()", v2.

Besides the obvious (and desired) difference between krealloc() and
kvrealloc(), there is some inconsistency in their function signatures and
behavior:

- krealloc() frees the memory when the requested size is zero, whereas
kvrealloc() simply returns a pointer to the existing allocation.

- krealloc() behaves like kmalloc() if a NULL pointer is passed, whereas
kvrealloc() does not accept a NULL pointer at all and, if passed, would fault
instead.

- krealloc() is self-contained, whereas kvrealloc() relies on the caller to
provide the size of the previous allocation.

Inconsistent behavior throughout allocation APIs is error prone, hence
make kvrealloc() behave like krealloc(), which seems superior in all
mentioned aspects.

In order to be able to get rid of kvrealloc()'s oldsize parameter,
introduce vrealloc() and make use of it in kvrealloc().

Making use of vrealloc() in kvrealloc() also provides oppertunities to
grow (and shrink) allocations more efficiently. For instance, vrealloc()
can be optimized to allocate and map additional pages to grow the
allocation or unmap and free unused pages to shrink the allocation.

Besides the above, those functions are required by Rust's allocator abstractons
[1] (rework based on this series in [2]). With `Vec` or `KVec` respectively,
potentially growing (and shrinking) data structures are rather common.

[1] https://lore.kernel.org/lkml/20240704170738.3621-1-dakr@redhat.com/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/dakr/linux.git/log/?h=rust/mm


This patch (of 2):

Implement vrealloc() analogous to krealloc().

Currently, krealloc() requires the caller to pass the size of the previous
memory allocation, which, instead, should be self-contained.

We attempt to fix this in a subsequent patch which, in order to do so,
requires vrealloc().

Besides that, we need realloc() functions for kernel allocators in Rust
too. With `Vec` or `KVec` respectively, potentially growing (and
shrinking) data structures are rather common.

[dakr@kernel.org: fix missing nommu implementation]
Link: https://lkml.kernel.org/r/20240725141227.13954-1-dakr@kernel.org
[dakr@kernel.org: document concurrency restrictions]
Link: https://lkml.kernel.org/r/20240725125442.4957-1-dakr@kernel.org
[dakr@kernel.org: consider spare memory for __GFP_ZERO]
Link: https://lkml.kernel.org/r/20240730185049.6244-3-dakr@kernel.org
[dakr@kernel.org: properly document __GFP_ZERO behavior]
Link: https://lkml.kernel.org/r/20240730185049.6244-4-dakr@kernel.org
Link: https://lkml.kernel.org/r/20240722163111.4766-1-dakr@kernel.org
Link: https://lkml.kernel.org/r/20240722163111.4766-2-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Chandan Babu R <chandan.babu@oracle.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: Oliver Upton <oliver.upton@linux.dev>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Wedson Almeida Filho <wedsonaf@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

authored by

Danilo Krummrich and committed by
Andrew Morton
3ddc2fef 5fe690a5

+79
+4
include/linux/vmalloc.h
··· 189 189 extern void *vcalloc_noprof(size_t n, size_t size) __alloc_size(1, 2); 190 190 #define vcalloc(...) alloc_hooks(vcalloc_noprof(__VA_ARGS__)) 191 191 192 + void * __must_check vrealloc_noprof(const void *p, size_t size, gfp_t flags) 193 + __realloc_size(2); 194 + #define vrealloc(...) alloc_hooks(vrealloc_noprof(__VA_ARGS__)) 195 + 192 196 extern void vfree(const void *addr); 193 197 extern void vfree_atomic(const void *addr); 194 198
+5
mm/nommu.c
··· 126 126 } 127 127 EXPORT_SYMBOL(__vmalloc_noprof); 128 128 129 + void *vrealloc_noprof(const void *p, size_t size, gfp_t flags) 130 + { 131 + return krealloc_noprof(p, size, (flags | __GFP_COMP) & ~__GFP_HIGHMEM); 132 + } 133 + 129 134 void *__vmalloc_node_range_noprof(unsigned long size, unsigned long align, 130 135 unsigned long start, unsigned long end, gfp_t gfp_mask, 131 136 pgprot_t prot, unsigned long vm_flags, int node,
+70
mm/vmalloc.c
··· 4030 4030 } 4031 4031 EXPORT_SYMBOL(vzalloc_node_noprof); 4032 4032 4033 + /** 4034 + * vrealloc - reallocate virtually contiguous memory; contents remain unchanged 4035 + * @p: object to reallocate memory for 4036 + * @size: the size to reallocate 4037 + * @flags: the flags for the page level allocator 4038 + * 4039 + * If @p is %NULL, vrealloc() behaves exactly like vmalloc(). If @size is 0 and 4040 + * @p is not a %NULL pointer, the object pointed to is freed. 4041 + * 4042 + * If __GFP_ZERO logic is requested, callers must ensure that, starting with the 4043 + * initial memory allocation, every subsequent call to this API for the same 4044 + * memory allocation is flagged with __GFP_ZERO. Otherwise, it is possible that 4045 + * __GFP_ZERO is not fully honored by this API. 4046 + * 4047 + * In any case, the contents of the object pointed to are preserved up to the 4048 + * lesser of the new and old sizes. 4049 + * 4050 + * This function must not be called concurrently with itself or vfree() for the 4051 + * same memory allocation. 4052 + * 4053 + * Return: pointer to the allocated memory; %NULL if @size is zero or in case of 4054 + * failure 4055 + */ 4056 + void *vrealloc_noprof(const void *p, size_t size, gfp_t flags) 4057 + { 4058 + size_t old_size = 0; 4059 + void *n; 4060 + 4061 + if (!size) { 4062 + vfree(p); 4063 + return NULL; 4064 + } 4065 + 4066 + if (p) { 4067 + struct vm_struct *vm; 4068 + 4069 + vm = find_vm_area(p); 4070 + if (unlikely(!vm)) { 4071 + WARN(1, "Trying to vrealloc() nonexistent vm area (%p)\n", p); 4072 + return NULL; 4073 + } 4074 + 4075 + old_size = get_vm_area_size(vm); 4076 + } 4077 + 4078 + /* 4079 + * TODO: Shrink the vm_area, i.e. unmap and free unused pages. What 4080 + * would be a good heuristic for when to shrink the vm_area? 4081 + */ 4082 + if (size <= old_size) { 4083 + /* Zero out spare memory. */ 4084 + if (want_init_on_alloc(flags)) 4085 + memset((void *)p + size, 0, old_size - size); 4086 + 4087 + return (void *)p; 4088 + } 4089 + 4090 + /* TODO: Grow the vm_area, i.e. allocate and map additional pages. */ 4091 + n = __vmalloc_noprof(size, flags); 4092 + if (!n) 4093 + return NULL; 4094 + 4095 + if (p) { 4096 + memcpy(n, p, old_size); 4097 + vfree(p); 4098 + } 4099 + 4100 + return n; 4101 + } 4102 + 4033 4103 #if defined(CONFIG_64BIT) && defined(CONFIG_ZONE_DMA32) 4034 4104 #define GFP_VMALLOC32 (GFP_DMA32 | GFP_KERNEL) 4035 4105 #elif defined(CONFIG_64BIT) && defined(CONFIG_ZONE_DMA)