rust: alloc: add missing trait item MIN_ALIGN to Cmalloc
Subsystems (3)
Danilo Krummrich
maintainerPATCH 1/1 rust: alloc: add missing trait item MIN_ALIGN to Cmalloc
Cmalloc is missing the trait item MIN_ALIGN introduced by commit 1b1a946dc2b5 ("rust: alloc: specify the minimum alignment of each allocator"), causing the following error on the rusttest make target.
error[E0046]: not all trait items implemented, missing: `MIN_ALIGN`
--> rust/kernel/alloc/allocator_test.rs:37:1
|
37 | unsafe impl Allocator for Cmalloc {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `MIN_ALIGN` in implementation
|
::: rust/kernel/alloc.rs:146:5
|
146 | const MIN_ALIGN: usize;
| ---------------------- `MIN_ALIGN` from traitImplement MIN_ALIGN for Cmalloc to fix this.
--- rust/kernel/alloc/allocator_test.rs | 2 ++ 1 file changed, 2 insertions(+)
rust/kernel/alloc/allocator_test.rs +2 -0
@@ -35,6 +35,8 @@// - passing a pointer to a valid memory allocation created by this `Allocator` is always OK, // - `realloc` provides the guarantees as provided in the `# Guarantees` section. unsafe impl Allocator for Cmalloc { +const MIN_ALIGN :usize = bindings::ARCH_KMALLOC_MINALIGN ; +unsafe fn realloc ( ptr:Option <NonNull <u8 >>, layout: Layout,
base-commit: ac9eea3d08c25fb213deb113d246ff5dadb31fbc
Signature · 1 line
2.50.1
Quoted 16 lines · Danilo Krummrich
Cmalloc is missing the trait item
MIN_ALIGNintroduced by commit1b1a946dc2b5("rust: alloc: specify the minimum alignment of each allocator"), causing the following error on therusttestmake target.error[E0046]: not all trait items implemented, missing: `MIN_ALIGN` --> rust/kernel/alloc/allocator_test.rs:37:1 | 37 | unsafe impl Allocator for Cmalloc { | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `MIN_ALIGN` in implementation | ::: rust/kernel/alloc.rs:146:5 | 146 | const MIN_ALIGN: usize; | ---------------------- `MIN_ALIGN` from trait
Quoted 6 lines
Implement
MIN_ALIGNfor Cmalloc to fix this.Reported-by Miguel Ojeda <[email protected]>Closes https://lore.kernel.org/all/CANiq72k0FSBTB2yOjiAy9PnAuyM=-PHxL3uQQ_Cv+zwswnr_bA@mail.gmail.com/ open hereFixes 1b1a946dc2b5 ("rust: alloc: specify the minimum alignment of each allocator")Signed-off-by Danilo Krummrich <[email protected]>
Thanks for taking of it. Assuming a satisfactory answer to the quesiton below:
rust/kernel/alloc/allocator_test.rs line 38
Quoted 13 lines
--- rust/kernel/alloc/allocator_test.rs | 2 ++ 1 file changed, 2 insertions(+)rust/kernel/alloc/allocator_test.rs +1
-0@@ -35,6 +35,8 @@// - passing a pointer to a valid memory allocation created by this `Allocator` is always OK, // - `realloc` provides the guarantees as provided in the `# Guarantees` section. unsafe impl Allocator for Cmalloc { +const MIN_ALIGN :usize = bindings::ARCH_KMALLOC_MINALIGN ;
Is this the right value for normal malloc?
Alice
Danilo Krummrich
maintainerQuoted 4 lines · Alice Ryhl
Quoted 1 line · Danilo Krummrich
+ const
MIN_ALIGN: usize =bindings::ARCH_KMALLOC_MINALIGN;Is this the right value for normal malloc?
Heh! ARCH_KMALLOC_MINALIGN should be correct, because the Cmalloc implementation should (ideally) enforce a minimum alignment of ARCH_KMALLOC_MINALIGN for compatibility reasons.
However, double checking the existing code, it doesn't. Hence, the correct value must either be
const MIN_ALIGN: usize = align_of::<crate::ffi::c_ulonglong>();
instead. Or, we have to actually enforce ARCH_KMALLOC_MINALIGN.
Given that this fix is only for within the -rc cycles, i.e. we will remove allocator_test.rs before the MIN_ALIGN stuff ever hits an actual release, just using align_of::<crate::ffi::c_ulonglong>() should be fine.
Danilo Krummrich
maintainerQuoted 13 lines · Danilo Krummrich
Quoted 4 lines · Alice Ryhl
Quoted 1 line · Danilo Krummrich
+ const
MIN_ALIGN: usize =bindings::ARCH_KMALLOC_MINALIGN;Is this the right value for normal malloc?
Heh!
ARCH_KMALLOC_MINALIGNshould be correct, because the Cmalloc implementation should (ideally) enforce a minimum alignment ofARCH_KMALLOC_MINALIGNfor compatibility reasons.However, double checking the existing code, it doesn't. Hence, the correct value must either be
Quoted 7 lines
const MIN_ALIGN: usize = align_of::<crate::ffi::c_ulonglong>();instead. Or, we have to actually enforce
ARCH_KMALLOC_MINALIGN.Given that this fix is only for within the -rc cycles, i.e. we will remove
allocator_test.rsbefore theMIN_ALIGNstuff ever hits an actual release, just using align_of::<crate::ffi::c_ulonglong>() should be fine.
Just for completeness, allocator_test.rs is also broken for the assumption that Vmalloc allocated memory is always PAGE_SIZE aligned.
Vmalloc::MIN_ALIGN, of course, does not report PAGE_SIZE in the case of allocator_test.rs being in charge, but it's an assumption that people might rightfully rely on in their (unsafe) code.
In the end, just another argument for getting rid of allocator_test.rs. :)
Danilo Krummrich
maintainerQuoted 1 line · Danilo Krummrich
In the end, just another argument for getting rid of
allocator_test.rs. :)
Actually, I think it's better to just do it:
Miguel, mind if I just apply [1] and [2]? (Just recognizing, probably this was your intention anyways?)
[1] https://lore.kernel.org/rust-for-linux/[email protected]/
[2] https://lore.kernel.org/rust-for-linux/[email protected]/
Quoted 5 lines · Danilo Krummrich
Actually, I think it's better to just do it:
Miguel, mind if I just apply [1] and [2]? (Just recognizing, probably this was your intention anyways?)
Yeah, since 1b1a946dc2b5 is in the same branch, and David already reviewed the main patch too, I think it is easier for you than trying to deal with all this otherwise.
(Whether applying or rebasing both are fine I would say, since rusttest is not a big deal and the chances of people bisecting on that are low.)
For the other fix I sent that you already applied through DRM, I think we did the right thing, because it was not alloc-next and it was fairly trivial, so it made sense.
Thanks!
Cheers,
Miguel
Danilo Krummrich
maintainerQuoted 3 lines · Miguel Ojeda
Yeah, since
1b1a946dc2b5is in the same branch, and David already reviewed the main patch too, I think it is easier for you than trying to deal with all this otherwise.
Yeah, I also think there's nothing to backport.
I.e. the fact that Cmalloc does not provide the same minimum alignment guarantees as Kmalloc and Vmalloc is not an issue for existing upstream code.
And fixing it for potential additions to the rusttest target in downstream code doesn't seem worth the effort.
Quoted 3 lines
For the other fix I sent that you already applied through DRM, I think we did the right thing, because it was not
alloc-nextand it was fairly trivial, so it made sense.
Yes, that one is correct.