Re: [PATCH v4 05/15] mm: introduce execmem_alloc() and execmem_free()

[Mike Rapoport]

On Mon, Apr 15, 2024 at 09:52:41AM +0200, Peter Zijlstra wrote:
> On Thu, Apr 11, 2024 at 07:00:41PM +0300, Mike Rapoport wrote:
> > +/**
> > + * enum execmem_type - types of executable memory ranges
> > + *
> > + * There are several subsystems that allocate executable memory.
> > + * Architectures define different restrictions on placement,
> > + * permissions, alignment and other parameters for memory that can be used
> > + * by these subsystems.
> > + * Types in this enum identify subsystems that allocate executable memory
> > + * and let architectures define parameters for ranges suitable for
> > + * allocations by each subsystem.
> > + *
> > + * @EXECMEM_DEFAULT: default parameters that would be used for types that
> > + * are not explcitly defined.
> > + * @EXECMEM_MODULE_TEXT: parameters for module text sections
> > + * @EXECMEM_KPROBES: parameters for kprobes
> > + * @EXECMEM_FTRACE: parameters for ftrace
> > + * @EXECMEM_BPF: parameters for BPF
> > + * @EXECMEM_TYPE_MAX:
> > + */
> > +enum execmem_type {
> > +	EXECMEM_DEFAULT,
> > +	EXECMEM_MODULE_TEXT = EXECMEM_DEFAULT,
> > +	EXECMEM_KPROBES,
> > +	EXECMEM_FTRACE,
> > +	EXECMEM_BPF,
> > +	EXECMEM_TYPE_MAX,
> > +};
> 
> Can we please get a break-down of how all these types are actually
> different from one another?
> 
> I'm thinking some platforms have a tiny immediate space (arm64 comes to
> mind) and has less strict placement constraints for some of them?

loongarch, mips, nios2 and sparc define modules address space different
from vmalloc and use that for modules, kprobes and bpf (where supported).

parisc uses vmalloc range for everything, but it sets permissions to
PAGE_KERNEL_RWX because it's PAGE_KERNEL_EXEC is read only and it lacks
set_memory_* APIs.

arm has an address space for modules, but it fall back to the entire
vmalloc with CONFIG_ARM_MODULE_PLTS=y.

arm64 uses different ranges for modules and bpf/kprobes. For kprobes it
does vmalloc(PAGE_KERNEL_ROX) and for bpf just plain vmalloc().
For modules arm64 first tries to allocated from 128M below kernel_end and
if that fails it uses 2G below kernel_end as a fallback.

powerpc uses vmalloc space for everything for some configurations. For
book3s-32 and 8xx it defines two ranges that are used for module text,
kprobes and bpf and the module data can be allocated anywhere in vmalloc.

riscv has an address space for modules, a different address space for bpf
and uses vmalloc space for kprobes.

s390 and x86 have modules address space and use that space for all
executable allocations.

The EXECMEM_FTRACE type is only used on s390 and x86 and for now it's there
more for completeness rather to denote special constraints or properties.

-- 
Sincerely yours,
Mike.