[patch][sh] Avoid reading uninitialized data in tc-sh.c

[Andrew Stubbs]

Hi all,

The attached patch corrects a defect in the the SH assembler, found 
using valgrind.

I believe the problem was benign - it was simply checking displacements 
for instructions that needed them, but before it had identified whether 
that instruction was relevant or not. If the random data was out of 
range for the displacement then the instruction was rejected, just as it 
would have been when it was found to be the wrong sort. Real problem or 
not, it's better to silence the warning so as not to hide other real 
problems.

The patch merely moves the displacement checking code until after the 
opcode matching code.

OK?

Andrew
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: binutils-valgrind.patch
URL: <https://sourceware.org/pipermail/binutils/attachments/20100219/f944ee32/attachment.ksh>