diff mbox

Fix unitialized warning on gdb/typeprint.c:whatis_exp

Message ID 20180116203239.27787-1-sergiodj@redhat.com
State New, archived
Headers

Commit Message

Sergio Durigan Junior Jan. 16, 2018, 8:32 p.m. UTC 
  This simple patch initializes "struct value *val" to NULL, which
silences a when compiling GDB with certain GCC versions.

This warning is technically incorrect, because there is now way that
"val" will be used unitialized if you look at the code flow, but it's
a simple "fix" and doesn't do any harm.

Is it OK to push this to master and 8.1?  I believe I will still need
to create a bug with the 8.1 milestone set, even for this simple
patch.

gdb/ChangeLog:
2017-01-16  Sergio Durigan Junior  <sergiodj@redhat.com>

	* typeprint.c (whatis_exp): Initialize "val" to NULL.
---
 gdb/typeprint.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Eli Zaretskii Jan. 17, 2018, 3:34 p.m. UTC | #1 
> From: Sergio Durigan Junior <sergiodj@redhat.com>
> Cc: Eli Zaretskii <eliz@gnu.org>,
> 	Sergio Durigan Junior <sergiodj@redhat.com>
> Date: Tue, 16 Jan 2018 15:32:39 -0500
> 
> This simple patch initializes "struct value *val" to NULL, which
> silences a when compiling GDB with certain GCC versions.
> 
> This warning is technically incorrect, because there is now way that
> "val" will be used unitialized if you look at the code flow, but it's
> a simple "fix" and doesn't do any harm.

This fixes the warning for me (of course).

Thanks.
Pedro Alves Jan. 17, 2018, 4:47 p.m. UTC | #2 
On 01/16/2018 08:32 PM, Sergio Durigan Junior wrote:
> This simple patch initializes "struct value *val" to NULL, which
> silences a when compiling GDB with certain GCC versions.

Please include a representative paste of warnings in git logs.
It makes it easier to see what a patch/commit is about.

> This warning is technically incorrect, because there is now way that
> "val" will be used unitialized if you look at the code flow, but it's
> a simple "fix" and doesn't do any harm.

No, it's not incorrect.  It's showing a real bug.  Try, e.g.:

(gdb) set print object on
(gdb) whatis some_structure_type

Thread 1 "gdb" received signal SIGSEGV, Segmentation fault.
0x00000000005dda90 in check_typedef (type=0x6120736573756170) at src/gdb/gdbtypes.c:2388
2388      int instance_flags = TYPE_INSTANCE_FLAGS (type);
(top-gdb) bt
#0  0x00000000005dda90 in check_typedef(type*) (type=0x6120736573756170) at src/gdb/gdbtypes.c:2388
#1  0x00000000005e63fb in gnuv3_rtti_type(value*, int*, LONGEST*, int*) (value=0xdf21b7, full_p=0x7fffffffd180, top_p=0x7fffffffd188, using_enc_p=0x7fffffffd184) at src/gdb/gnu-v3-abi.c:293
#2  0x000000000055f7d8 in value_rtti_type(value*, int*, long*, int*) (v=0xdf21b7, full=0x7fffffffd180, top=0x7fffffffd188, using_enc=0x7fffffffd184) at src/gdb/cp-abi.c:117
#3  0x00000000006cc61f in whatis_exp(char const*, int) (exp=<optimized out>, show=-1) at src/gdb/typeprint.c:515
#4  0x0000000000472462 in cmd_func(cmd_list_element*, char const*, int) (cmd=<optimized out>, args=<optimized out>, from_tty=<optimized out>)
    at src/gdb/cli/cli-decode.c:1886
#5  0x00000000006b833a in execute_command(char const*, int) (p=<optimized out>, from_tty=1) at src/gdb/top.c:630
#6  0x00000000005c0f8c in command_handler(char const*) (command=0xdf21b0 "whatis siginfo_t") at src/gdb/event-top.c:583
#7  0x00000000005c12d8 in command_line_handler(char*) (rl=<optimized out>) at /home/pedro/gdb/mygit/src/gdb/event-top.c:774
...

Thanks,
Pedro Alves
Sergio Durigan Junior Jan. 17, 2018, 6:03 p.m. UTC | #3 
On Wednesday, January 17 2018, Pedro Alves wrote:

> On 01/16/2018 08:32 PM, Sergio Durigan Junior wrote:
>> This simple patch initializes "struct value *val" to NULL, which
>> silences a when compiling GDB with certain GCC versions.
>
> Please include a representative paste of warnings in git logs.
> It makes it easier to see what a patch/commit is about.

Will do.

>> This warning is technically incorrect, because there is now way that
>> "val" will be used unitialized if you look at the code flow, but it's
>> a simple "fix" and doesn't do any harm.
>
> No, it's not incorrect.  It's showing a real bug.  Try, e.g.:
>
> (gdb) set print object on
> (gdb) whatis some_structure_type
>
> Thread 1 "gdb" received signal SIGSEGV, Segmentation fault.
> 0x00000000005dda90 in check_typedef (type=0x6120736573756170) at src/gdb/gdbtypes.c:2388
> 2388      int instance_flags = TYPE_INSTANCE_FLAGS (type);
> (top-gdb) bt
> #0  0x00000000005dda90 in check_typedef(type*) (type=0x6120736573756170) at src/gdb/gdbtypes.c:2388
> #1 0x00000000005e63fb in gnuv3_rtti_type(value*, int*, LONGEST*, int*)
> (value=0xdf21b7, full_p=0x7fffffffd180, top_p=0x7fffffffd188,
> using_enc_p=0x7fffffffd184) at src/gdb/gnu-v3-abi.c:293
> #2  0x000000000055f7d8 in value_rtti_type(value*, int*, long*, int*) (v=0xdf21b7, full=0x7fffffffd180, top=0x7fffffffd188, using_enc=0x7fffffffd184) at src/gdb/cp-abi.c:117
> #3  0x00000000006cc61f in whatis_exp(char const*, int) (exp=<optimized out>, show=-1) at src/gdb/typeprint.c:515
> #4  0x0000000000472462 in cmd_func(cmd_list_element*, char const*, int) (cmd=<optimized out>, args=<optimized out>, from_tty=<optimized out>)
>     at src/gdb/cli/cli-decode.c:1886
> #5  0x00000000006b833a in execute_command(char const*, int) (p=<optimized out>, from_tty=1) at src/gdb/top.c:630
> #6  0x00000000005c0f8c in command_handler(char const*) (command=0xdf21b0 "whatis siginfo_t") at src/gdb/event-top.c:583
> #7  0x00000000005c12d8 in command_line_handler(char*) (rl=<optimized out>) at /home/pedro/gdb/mygit/src/gdb/event-top.c:774
> ...

Ah, I totally missed that.  Sorry about jumping into the conclusion that
it was incorrect.

I'll try to investigate what's happening here.
diff mbox

Patch

diff --git a/gdb/typeprint.c b/gdb/typeprint.c
index 9a125076a1..bf9aec5436 100644
--- a/gdb/typeprint.c
+++ b/gdb/typeprint.c
@@ -405,7 +405,7 @@  error_unknown_type (const char *sym_print_name)
 static void
 whatis_exp (const char *exp, int show)
 {
-  struct value *val;
+  struct value *val = NULL;
   struct cleanup *old_chain;
   struct type *real_type = NULL;
   struct type *type;