Skip to contentSkip to conversation

x86_64: Eliminate atleast 10684 sparse warnings

PATCH Applied Jaswinder Singh Rajput 3 participants 4 messages Dec 27, 2008

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86_64: Eliminate atleast 10684 sparse warnings

Impact: cleanup, avoid 10000+ sparse warnings

Eliminate 1434 sparse warnings like:
include/linux/list.h:106:16: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1434 sparse warnings like:
include/linux/list.h:107:16: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1434 sparse warnings like:
include/linux/list.h:579:12: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1434 sparse warnings like:
include/linux/list.h:580:13: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1237 sparse warnings like:
include/linux/rculist.h:97:16: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1237 sparse warnings like:
include/linux/rculist.h:143:14: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1237 sparse warnings like:
include/linux/rculist.h:260:13: warning: constant 0xdead000000000000 is so big it is unsigned long

Eliminate 1237 sparse warnings like:
include/linux/rculist.h:280:15: warning: constant 0xdead000000000000 is so big it is unsigned long

---
 include/linux/poison.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
include/linux/poison.h +1 -1
index 603d3bd..f2de9a9 100644
@@ -9,7 +9,7 @@
  * that is also not mappable by user-space exploits:
  */
 #ifdef CONFIG_ILLEGAL_POINTER_VALUE
-# define POISON_POINTER_DELTA CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
 #else
 # define POISON_POINTER_DELTA 0
 #endif
Signature · 1 line
1.5.5.1
Applied
#1 lore raw
Quoted 1 line · Jaswinder Singh Rajput

Impact: cleanup, avoid 10000+ sparse warnings

:-)

Quoted 2 lines
 include/linux/poison.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

applied to tip/safe-poison-pointers, thanks Jaswinder!

Ingo

#2 lore raw
In reply to Jaswinder Singh Rajput

include/linux/poison.h line 15

Quoted 6 lines · On Sat, 27 Dec 2008 15:21:36 +0530 Jaswinder Singh Rajput…

#ifdef CONFIG_ILLEGAL_POINTER_VALUE

include/linux/poison.h +1 -1
-# define POISON_POINTER_DELTA CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
 #else
 # define POISON_POINTER_DELTA 0
 #endif

So now

commit 948f984df52511bb0efa5c026813b0c34de43aa0

Author:     Avi Kivity <[email protected]>
AuthorDate: Tue May 20 14:39:25 2008 +0300
Commit:     Ingo Molnar <[email protected]>
CommitDate: Tue May 20 13:42:31 2008 +0200

core, x86: make LIST_POISON less deadly

gets splattered across three commits (at least). Sigh.

(how come that patch is 7 moonths old and still unmerged, btw?)

#3 lore raw
Quoted 12 lines · Andrew Morton

include/linux/poison.h line 15

Quoted 6 lines · On Sat, 27 Dec 2008 15:21:36 +0530 Jaswinder Singh Rajput…

#ifdef CONFIG_ILLEGAL_POINTER_VALUE

include/linux/poison.h +1 -1
-# define POISON_POINTER_DELTA CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
 #else
 # define POISON_POINTER_DELTA 0
 #endif

So now

commit 948f984df52511bb0efa5c026813b0c34de43aa0

Quoted 8 lines
Author:     Avi Kivity <[email protected]>
AuthorDate: Tue May 20 14:39:25 2008 +0300
Commit:     Ingo Molnar <[email protected]>
CommitDate: Tue May 20 13:42:31 2008 +0200

core, x86: make LIST_POISON less deadly

gets splattered across three commits (at least). Sigh.

Can rebase, although that would hide the (deserved) credit that Jaswinder's patch deserves. And it does not appear to be a must-rebase example to me - it's not like one does sparse builds during bisection.

Quoted 1 line

(how come that patch is 7 moonths old and still unmerged, btw?)

Linus had reservations about the first version (which got fixed but which prevented it from going upstream in the first cycle) - and it's not an exactly critical change and other stuff interfered. All our previous merge windows were fully filled ;)

Ingo