* [OpenRISC] architectures that still need to remove set_fs()
@ 2022-01-18 7:48 Christoph Hellwig
2022-01-18 11:48 ` Stafford Horne
0 siblings, 1 reply; 2+ messages in thread
From: Christoph Hellwig @ 2022-01-18 7:48 UTC (permalink / raw)
To: openrisc
Hi all,
you are in this list because your architecture still implements and
uses address space overrides using set_fs(), which are deprecated and
have been removed from all mainstream architecture ports. To help
cleanup the core kernel it would be great to make progress on removing
set_fs entirely.
The following steps are required:
(1) implement the __get_kernel_nofault and __put_kernel_nofault
helper to access kernel memory without page faults, replacing
the get/put_user under set_fs(KERNEL_DS) abuse. Mips has a good
example for a trivial implementation for architectures that use
a common address space in commit 04324f44cb69
(2) remove any architecture specific use of such constructs. This
only affects ia64 and sh.
(3) stop selecting the SET_FS and remove all the set_fs-related
infrastructure. The above mips commit is a good guide once again.
Thanks!
^ permalink raw reply [flat|nested] 2+ messages in thread
* [OpenRISC] architectures that still need to remove set_fs()
2022-01-18 7:48 [OpenRISC] architectures that still need to remove set_fs() Christoph Hellwig
@ 2022-01-18 11:48 ` Stafford Horne
0 siblings, 0 replies; 2+ messages in thread
From: Stafford Horne @ 2022-01-18 11:48 UTC (permalink / raw)
To: openrisc
Hi,
On Tue, Jan 18, 2022 at 08:48:36AM +0100, Christoph Hellwig wrote:
> Hi all,
>
> you are in this list because your architecture still implements and
> uses address space overrides using set_fs(), which are deprecated and
> have been removed from all mainstream architecture ports. To help
> cleanup the core kernel it would be great to make progress on removing
> set_fs entirely.
>
> The following steps are required:
>
> (1) implement the __get_kernel_nofault and __put_kernel_nofault
> helper to access kernel memory without page faults, replacing
> the get/put_user under set_fs(KERNEL_DS) abuse. Mips has a good
> example for a trivial implementation for architectures that use
> a common address space in commit 04324f44cb69
Also see commit 08ee3a009f49a which is preparation for removing set_fs/get_fs
this commit implements __get_kernel_nofault/__put_kernel_nofault for mips.
> (2) remove any architecture specific use of such constructs. This
> only affects ia64 and sh.
> (3) stop selecting the SET_FS and remove all the set_fs-related
> infrastructure. The above mips commit is a good guide once again.
>
> Thanks!
I will look into this for OpenRISC.
-Stafford
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-01-18 11:48 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2022-01-18 7:48 [OpenRISC] architectures that still need to remove set_fs() Christoph Hellwig
2022-01-18 11:48 ` Stafford Horne
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox