Re: c/2371: Hang with long long function argument.

Re: c/2371: Hang with long long function argument.

[Dave Korn]

The following reply was made to PR c/2371; it has been noted by GNATS.

From: "Dave Korn" <davek-ml@ntlworld.com>
To: <snowball3@bigfoot.com>,
	<gcc-gnats@gcc.gnu.org>
Cc:  
Subject: Re: c/2371: Hang with long long function argument.
Date: Sun, 1 Apr 2001 06:45:55 +0100

 ----- Original Message -----
 From: <snowball3@bigfoot.com>
 To: <gcc-gnats@gcc.gnu.org>
 Sent: Saturday, March 24, 2001 4:53 AM
 Subject: c/2371: Hang with long long function argument.
 
 
 > >Description:
 > It seems that code generated for handling variables of
 > type 'long long' is incorrect. The sample program when
 > run hangs. Changing the type to 'long' allows the program
 > to exit normally.
 > >How-To-Repeat:
 > # 1 "longlong.c"
 > int long_long_test(  long long x)
 > {
 >   do
 >   {
 >     x =- 1;
 >   } while (x);
 
   It seems that you meant to say x-=1 there.  Otherwise looping forever is
 the expected behaviour.  I've tested your code with both longs and long
 longs, and both with and without your bugfix.  Both the bugged versions
 looped forever and both the fixed versions exit almost instantly.  Are you
 sure you didn't fix the bug at the same time as you changed from long longs
 to plain longs?
 
          DaveK
 
 
 

Re: c/2371: Hang with long long function argument.

[Mark E.]

The following reply was made to PR c/2371; it has been noted by GNATS.

From: "Mark E." <snowball3@bigfoot.com>
To: "Dave Korn" <davek-ml@ntlworld.com>,
              <gcc-gnats@gcc.gnu.org>
Cc:  
Subject: Re: c/2371: Hang with long long function argument.
Date: Sun, 1 Apr 2001 10:13:55 -0400

 > looped forever and both the fixed versions exit almost instantly.  Are you
 > sure you didn't fix the bug at the same time as you changed from long longs
 > to plain longs?
 
 After going through the assembly code, it turned out the code that prompted 
 the testcase was using the wrong prototype causing the wrong number of 
 arguments to be pushed on to the stack. The unfortunate typo delayed me in 
 finding the real problem. When the testcase also hung, I thought for sure I 
 was on to something. But when I looked at its assembly and noticed it was 
 setting x to -1LL, I realized the testcase was faulty and had it closed. 
 Sorry for any trouble. It certainly caused me some.
 
 Mark
 
 
 

c/2371: Hang with long long function argument.

[snowball3]