mbox series

[0/3] Fixes for usercopy

Message ID 20220612213227.3881769-1-willy@infradead.org (mailing list archive)
Headers show
Series Fixes for usercopy | expand

Message

Matthew Wilcox (Oracle) June 12, 2022, 9:32 p.m. UTC
Kees, I'm hoping you'll take these through your tree.  I think they're
all reasonable fixes to go into 5.19.  The first one is essential;
it fixes two different bugs that people have hit.

Matthew Wilcox (Oracle) (3):
  usercopy: Handle vm_map_ram() areas
  usercopy: Cast pointer to an integer once
  usercopy: Make usercopy resilient against ridiculously large copies

 include/linux/vmalloc.h |  1 +
 mm/usercopy.c           | 24 +++++++++++++-----------
 mm/vmalloc.c            |  2 +-
 3 files changed, 15 insertions(+), 12 deletions(-)

Comments

Zorro Lang June 13, 2022, 8:04 a.m. UTC | #1
On Sun, Jun 12, 2022 at 10:32:24PM +0100, Matthew Wilcox (Oracle) wrote:
> Kees, I'm hoping you'll take these through your tree.  I think they're
> all reasonable fixes to go into 5.19.  The first one is essential;
> it fixes two different bugs that people have hit.
> 
> Matthew Wilcox (Oracle) (3):
>   usercopy: Handle vm_map_ram() areas
>   usercopy: Cast pointer to an integer once
>   usercopy: Make usercopy resilient against ridiculously large copies

Hi Matthew,

[Quick response] ...
After merging this patchset onto linux v5.19-rc2, I can't reproduce bug [1]
anymore. I'd like to keep the test running, if it find other issues I'll
feedback.

Thanks,
Zorro

[1]
https://bugzilla.kernel.org/show_bug.cgi?id=216073

> 
>  include/linux/vmalloc.h |  1 +
>  mm/usercopy.c           | 24 +++++++++++++-----------
>  mm/vmalloc.c            |  2 +-
>  3 files changed, 15 insertions(+), 12 deletions(-)
> 
> -- 
> 2.35.1
>
Kees Cook June 13, 2022, 4:25 p.m. UTC | #2
On Sun, 12 Jun 2022 22:32:24 +0100, Matthew Wilcox (Oracle) wrote:
> Kees, I'm hoping you'll take these through your tree.  I think they're
> all reasonable fixes to go into 5.19.  The first one is essential;
> it fixes two different bugs that people have hit.
> 
> Matthew Wilcox (Oracle) (3):
>   usercopy: Handle vm_map_ram() areas
>   usercopy: Cast pointer to an integer once
>   usercopy: Make usercopy resilient against ridiculously large copies
> 
> [...]

Applied to for-next/hardening, thanks!

[1/3] usercopy: Handle vm_map_ram() areas
      https://git.kernel.org/kees/c/751ad8bdde7f
[2/3] usercopy: Cast pointer to an integer once
      https://git.kernel.org/kees/c/de2ae8f5331a
[3/3] usercopy: Make usercopy resilient against ridiculously large copies
      https://git.kernel.org/kees/c/630b2014e60e