| 1 |
From 49beadbd47c270a00754c107a837b4f29df4c822 Mon Sep 17 00:00:00 2001 |
| 2 |
Message-ID: <49beadbd47c270a00754c107a837b4f29df4c822.1744656535.git.osandov@fb.com> |
| 3 |
From: Linus Torvalds <[email protected]> |
| 4 |
Date: Thu, 9 Jun 2022 09:41:42 -0700 |
| 5 |
Subject: [PATCH] gcc-12: disable '-Wdangling-pointer' warning for now |
| 6 |
MIME-Version: 1.0 |
| 7 |
Content-Type: text/plain; charset=UTF-8 |
| 8 |
Content-Transfer-Encoding: 8bit |
| 9 |
|
| 10 |
While the concept of checking for dangling pointers to local variables |
| 11 |
at function exit is really interesting, the gcc-12 implementation is not |
| 12 |
compatible with reality, and results in false positives. |
| 13 |
|
| 14 |
For example, gcc sees us putting things on a local list head allocated |
| 15 |
on the stack, which involves exactly those kinds of pointers to the |
| 16 |
local stack entry: |
| 17 |
|
| 18 |
In function ‘__list_add’, |
| 19 |
inlined from ‘list_add_tail’ at include/linux/list.h:102:2, |
| 20 |
inlined from ‘rebuild_snap_realms’ at fs/ceph/snap.c:434:2: |
| 21 |
include/linux/list.h:74:19: warning: storing the address of local variable ‘realm_queue’ in ‘*&realm_27(D)->rebuild_item.prev’ [-Wdangling-pointer=] |
| 22 |
74 | new->prev = prev; |
| 23 |
| ~~~~~~~~~~^~~~~~ |
| 24 |
|
| 25 |
But then gcc - understandably - doesn't really understand the big |
| 26 |
picture how the doubly linked list works, so doesn't see how we then end |
| 27 |
up emptying said list head in a loop and the pointer we added has been |
| 28 |
removed. |
| 29 |
|
| 30 |
Gcc also complains about us (intentionally) using this as a way to store |
| 31 |
a kind of fake stack trace, eg |
| 32 |
|
| 33 |
drivers/acpi/acpica/utdebug.c:40:38: warning: storing the address of local variable ‘current_sp’ in ‘acpi_gbl_entry_stack_pointer’ [-Wdangling-pointer=] |
| 34 |
40 | acpi_gbl_entry_stack_pointer = ¤t_sp; |
| 35 |
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~ |
| 36 |
|
| 37 |
which is entirely reasonable from a compiler standpoint, and we may want |
| 38 |
to change those kinds of patterns, but not not. |
| 39 |
|
| 40 |
So this is one of those "it would be lovely if the compiler were to |
| 41 |
complain about us leaving dangling pointers to the stack", but not this |
| 42 |
way. |
| 43 |
|
| 44 |
Signed-off-by: Linus Torvalds <[email protected]> |
| 45 |
--- |
| 46 |
Makefile | 3 +++ |
| 47 |
1 file changed, 3 insertions(+) |
| 48 |
|
| 49 |
diff --git a/Makefile b/Makefile |
| 50 |
index c43d825a3c4c..09208ffca353 100644 |
| 51 |
--- a/Makefile |
| 52 |
+++ b/Makefile |
| 53 |
@@ -805,6 +805,9 @@ endif |
| 54 |
KBUILD_CFLAGS += $(call cc-disable-warning, unused-but-set-variable) |
| 55 |
KBUILD_CFLAGS += $(call cc-disable-warning, unused-const-variable) |
| 56 |
|
| 57 |
+# These result in bogus false positives |
| 58 |
+KBUILD_CFLAGS += $(call cc-disable-warning, dangling-pointer) |
| 59 |
+ |
| 60 |
ifdef CONFIG_FRAME_POINTER |
| 61 |
KBUILD_CFLAGS += -fno-omit-frame-pointer -fno-optimize-sibling-calls |
| 62 |
else |
| 63 |
-- |
| 64 |
2.49.0 |
| 65 |
|