xen: bug fixes for 4.7-rc0

GIT PULL David Vrabel 2 participants 3 messages May 24, 2016 – Jun 22, 2016

lore

Share & tools

lore

David Vrabel

author

2016-05-24 14:56 UTC #0 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Linus,

Please git pull the following tag:

git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip.git for-linus-4.7-rc0-tag

xen: bug fixes for 4.7-rc0

Thanks.

David

 arch/x86/pci/xen.c               |  7 +++--
 arch/x86/xen/setup.c             | 65 ++++++++++++++++------------------------
 arch/x86/xen/time.c              |  6 ++--
 drivers/xen/Makefile             |  1 +
 drivers/xen/events/events_base.c |  6 ++--
 drivers/xen/gntdev.c             |  2 +-
 6 files changed, 40 insertions(+), 47 deletions(-)

Arnd Bergmann (1):
Xen: don't warn about 2-byte wchar_t in efi

David Vrabel (1):
xen/gntdev: reduce copy batch size to 16

Juergen Gross (1):
xen: use same main loop for counting and remapping pages

Ross Lagerwall (1):
xen/events: Don't move disabled irqs

Stefano Stabellini (2):
xen/x86: don't lose event interrupts
xen/x86: actually allocate legacy interrupts on PV guests
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJXRGuCAAoJEFxbo/MsZsTRqn0H/iDtaJVIWjSwXq8BuJzU2f8d CYJITZVZ67tkXuQMF3Ol+Mf9Gf8cG1PeSA6RsO1o/cFPAtwmhTW97VYw7dcfSTn5 t5skul1rKkrx9bFLaSO8IeExycIGdAYQUZRVoqRh4dwTXqDJeuPmFcEF515HkmaE AkdXc0xe1b9pppzc3kqG2kBIk6l/CgIUYZY7NsIlQxLRkHm0BPmtySsypYreXo2W /53mixS39g80IUIjnK0jRpg4o3T8FY+HKqQFP64kIDq6r0IfRXDbx/DY2OeUsTeW LTUH5/6d3y+oPN2WoOyZLAeYCzr7saheLAYouqpIs4o8yMV0s9IYEk7jTiTZIlc= =kFQS
-----END PGP SIGNATURE-----

Linus Torvalds

2016-06-22 21:25 UTC #1 lore raw

Having upgraded one of my machines to F24, I get a few new warnings during the kernel compile due to a new compiler.

Some of them are just annoying and wrong, but one of them points to a real Xen bug:

arch/x86/xen/mmu.c:1116:57: warning: array subscript is above array
bounds [-Warray-bounds]

    for (; vaddr <= vaddr_end && (pmd < (level2_kernel_pgt + PAGE_SIZE));
                                        ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~

because that is definitely completely wrong.

Yes, level2_kernel_pgt is one page in size, but it only has 512 entries, because each entry is 8 bytes.

So that "+ PAGE_SIZE" is entirely bogus. Either it should be

(level2_kernel_pgt + 512)

or it should be

((void *)level2_kernel_pgt + PAGE_SIZE)

but as it stands now it's a bug.

This harkens back to 2012, commit 7f9140626c757 ("xen/mmu: Copy and revector the P2M tree").

It may be that we end up never actually crossing the pmd boundary anyway due to vaddr limit - I didn't check. But please fix the code regardless.

Linus