Skip to content

/proc/kmem fixes and hwpoison bits v3

Wu Fengguang lkml 1 version New

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest4 patches 0 repliesNew

Changes since v2 from the cover letter

- remove the prototype change and make the bug fix simple
- add hwpoison checks for /dev/mem (this is in fact an old patch for me)

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/4devmem: fix kmem write bug on memory holesdrivers/char/mem.c+1 -10 replies
2/4devmem: check vmalloc address on kmem read/writedrivers/char/mem.c+16 -80 replies
3/4HWPOISON: prevent /dev/kmem users from accessing hwpoison pagesdrivers/char/mem.c+14 -40 replies
4/4HWPOISON: stop /dev/mem users from accessing hwpoison pagesdrivers/char/mem.c+20 -170 replies

Reviews trailers per patch

2 of 4 patches have a Reviewed-by or Acked-by.

PatchKAMEZAWA HiroyukiAndi Kleen
1/4
2/4
3/4R-b
4/4R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/4] /proc/kmem fixes and hwpoison bits v3

Hi,

Here are 2 bug fixes and also hwpoison checks for /dev/mem and /dev/kmem. The bug fixes are intended for this merge window.

Changes since v2:
- remove the prototype change and make the bug fix simple
- add hwpoison checks for /dev/mem (this is in fact an old patch for me)

Changes since v1:
- change vread()/vwrite() prototype (proposed by Kame)
- include Kame's is_vmalloc_or_module_addr() check and use Hugh's -ENXIO.
- I decided not to use __GFP_ZERO, since the buf will be reused in the loop, so it would be better for vread to zero-fill it each time.
- removed the hwpoison checks for vmalloc pages. It seems that vread/vwrite could be simplified to handle one single page, and the hwpoison bits can be considered after that.

Thanks,
Fengguang
--