AFAICS, fallout from repacing include of blkdev.h with include of bio.h.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
This fixes a lost 'key' variable declaration that went missing in a
mismerge (commit b981d8b3f5)
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
if your mask is host-endian, you should apply it after le64_to_cpu();
if it's little-endian - before. Doing both (for the same mask and
little-endian value) is broken.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
all uses of and almost all assignments to lro_desc->tcp_ack assume that it's
net-endian; one converts net-endian to host-endian and sticks it in
lro_desc->tcp_ack.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
deal with signedness of the stuff passed to set_bit() et.al.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
... since that sucker is not 32bit-only and on 64bit skb->tail is an
offset, not a pointer.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Using readw() and friends => needs to pull io.h and not all targets are
doing that via indirect chains.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
ohci_bus_{suspend,resume} exists only if we have CONFIG_PM; do the same
thing as other subdrivers...
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
In commit 4665079cbb ("[NETNS]: Move some
code into __init section when CONFIG_NET_NS=n") we got a new section -
.exit.text.refok (more of 'let's tell modpost that some bogus calls are
not bogus', a-la text.init.refok).
Unfortunately, the commit in question forgot to add it to TEXT_TEXT,
with rather amusing results.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Breaks on any target that has copy_to_user() defined as a non-trivial
macro.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
a) include/asm-um/arch can't just point to include/asm-$(SUBARCH) now
b) arch/{i386,x86_64}/crypto are merged now
c) subarch-obj needed changes
d) cpufeature_64.h should pull "cpufeature_32.h", not <asm/cpufeature_32.h>
since it can be included from asm-um/cpufeature.h
e) in case of uml-i386 we need CONFIG_X86_32 for make and gcc, but not
for Kconfig
f) sysctl.c shouldn't do vdso_enabled for uml-i386 (actually, that one
should be registered from corresponding arch/*/kernel/*, with ifdef
going away; that's a separate patch, though).
With that and with Stephen's patch ("[PATCH net-2.6] uml: hard_header fix")
we have uml allmodconfig building both on i386 and amd64.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Async signals should not be reported as sent by current in audit log. As
it is, we call audit_signal_info() too early in check_kill_permission().
Note that check_kill_permission() has that test already - it needs to know
if it should apply current-based permission checks. So the solution is to
move the call of audit_signal_info() between those.
Bogosity in question is easily reproduced - add a rule watching for e.g.
kill(2) from specific process (so that audit_signal_info() would not
short-circuit to nothing), say load_policy, watch the bogus OBJ_PID entry
in audit logs claiming that write(2) on selinuxfs file issued by
load_policy(8) had somehow managed to send a signal to syslogd...
Signed-off-by: Al Viro <[email protected]>
Acked-by: Steve Grubb <[email protected]>
Acked-by: Eric Paris <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
It gets pointer to fastcall function, expects a pointer to normal
one and calls the sucker.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
drivers/acpi/ec.c: In function `acpi_ec_ecdt_probe':
drivers/acpi/ec.c:873: warning: passing arg 1 of `acpi_get_devices' discards qualifiers from pointer target type
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Len Brown <[email protected]>
The check for audit_signals is misplaced and the check for
audit_dummy_context() is missing; as the result, if we send a signal to
auditd from task with NULL ->audit_context while we have audit_signals
!= 0 we end up with an oops.
Signed-off-by: Al Viro <[email protected]>
Acked-by: James Morris <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
C99 6.10.3[11]: preprocessing directive within the argument list of
macro invocation => undefined behaviour. Don't do that...
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
* wrong argument passed to pci_unmap_single() on failure
exit paths
* leak in the same area
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
#x blocks expansion of macro argument, but it won't do you any
good if it's already been expanded... As it is, RFALSE(cond, ....)
ended up with stringified _expanded_ cond. Real fun when cond contains
something like le32_to_cpu() and you are on a big-endian box...
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
... because somebody had added preempt.h -> list.h, resulting in
asm/system.h -> hardirq.h -> preempt.h -> list.h -> asm/system.h on m68k,
with smp_wmb() used in list.h and defined in asm/system.h below the include
of hardirq.h.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Have put_unaligned() warn if types would be wrong
for assignment, slap force-casts where needed. Cast the
result of get_unaligned to typeof(*ptr). With that in
place we get proper typechecking, both from gcc and from sparse,
including that for bitwise types.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Since powerpc insists on printing the _value_ of condition
and on casting it to long... At least let's make it a force-cast.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
We really need force-cast when converting to final result type;
unsigned long can be silently converted to integer types and
to pointers, but not to bitwise.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Address of auto variable is not a userland pointer. A good thing, too,
since if pppol2tp_tunnel_getsockopt() would _really_ get a userland pointer
as argument, it would be an instant roothole...
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
u32* volatile cyclone_timer means volatile auto pointer to u32,
which is clearly not what had been intended (we never even take
the address of that variable, let alone pass it to something that
could change it behind our back). u32 volatile * is what the
authors apparently wanted to say, but in reality we don't need that
qualifier there at all - it's (properly) only passed to iomem helpers
which takes care of that stuff just fine.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Convert rel_info to host-endian before calling ip6_tnl_err().
The things become much more straightforward that way.
The key observation (and the reason why that code actually
worked) is that after ip6_tnl_err() we either immediately
bailed out or had rel_info set to 0 or had it set to host-endian
and guaranteed to hit
(rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED)
case. So inconsistent endianness didn't really lead to bugs,
but it had been subtle and prone to breakage. New variant is
saner and obviously safe.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
struct bin_attribute * is needed in bin_attribute ->read()/->write()
now. Incidentally, could people please run the fscking compiler
before and after applying their patch and compare the build logs?
That (and many, many other) would be caught immediately.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Fallout from f8a7c6fe14. However, looking
at it shows that checks done in ASUS_LED_UNREGISTER() can't trigger
at all (we never get to asus_led_exit() if registration fails) and
if that registration fails, we actually leak stuff. IOW, it's worse
than just replacing class_dev with dev in there - the tests themselves
had been papering over the lousy cleanup logics.
Signed-off-by: Al Viro <[email protected]>
Michal Piotrowski <[email protected]>
Cc: Corentin Chary <[email protected]>
Cc: Karol Kozimor <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
> Subject : drivers/misc/asus-laptop.c:*: error: 'struct led_classdev' has no member named 'class_dev'
> References : http://lkml.org/lkml/2007/7/22/299
> Submitter : Gabriel C <[email protected]>
Fallout from f8a7c6fe14. However, looking
at it shows that checks done in ASUS_LED_UNREGISTER() can't trigger
at all (we never get to asus_led_exit() if registration fails) and
if that registration fails, we actually leak stuff. IOW, it's worse
than just replacing class_dev with dev in there - the tests themselves
had been papering over the lousy cleanup logics.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Len Brown <[email protected]>
Now that the last inlined instances are gone, all that is left to do
is turning disable_irq_nosync on arm26 and m68k from defines to aliases
and we are all set - we can make these externs in linux/interrupt.h
uncoditional and kill remaining instances in asm/irq.h
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Obviously broken on little-endian; fortunately, the option is not
frequently used...
Signed-off-by: Al Viro <[email protected]>
[ Hey, sparse is wonderful, but even better than sparse is having people
like Al that actually _run_ it and fix bugs using it. - Linus ]
Signed-off-by: Linus Torvalds <[email protected]>
Fix a mismerge in commit 8b6f50ef1d:
"spufs: make signal-notification files readonly for NOSCHED contexts",
where structs got duplicated.
Signed-off-by: Al Viro <[email protected]>
Acked-by: Jeremy Kerr <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Since we have use like ~SLUB_DMA, we ought to have the type
set right in both cases.
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Doing |= 1 << 19 to 16bit unsigned is not particulary useful;
that register is 32bit, unlike the ones dealt with in the rest of
function, so we need u32 variable here.
Signed-off-by: Al Viro <[email protected]>
Acked-by: Stephen Hemminger <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
anything that wants working dma-mapping won't work
parport_pc won't work on m68k unless we have ISA
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
i.e. tell modpost that entry point code (that has to be outside
of .init.text for external reasons) is OK to refer to .init.*
Signed-off-by: Al Viro <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Al Viro