In a message dated 7/17/2005 7:03:57 P.M. Central Daylight Time, [EMAIL PROTECTED] writes:
5 is always less than 4095. DUH. Thanks for pointing that out. And 4094 is also always less than 4095. The poster used 5 as an example. He could also have used 4094 or 4095 as an example, in which case my possibly unlikely reason for coding =H'xxxx' rather than LA Ry,xxxx would be a little more believable. My point was not how close 5 is to 4095, but rather the difficulty in patching an instruction when the patch only allows 12 bits to be changed (in an LA instruction) vs. when the patch allows 16 bits to be changed (when changing a half word literal). And LHI is not available as a solution when you are writing code that might execute on non-z/Arch. processors, of which there are still just a few in existence. >If you found it necessary to fix a problem by increasing the value >being loaded into the register to a number larger than 4095 but >could not easily change the source code and reassemble, you could >possibly patch the half word in the literal pool That's high risk behavior. I don't understand this reply. What is the risk? On the one hand, you have a LA instruction, perhaps containing an original value of 5 (or 4095) which for the purpose of fixing the bug must now be patched to contain 4096 and which is impossible, vs. a LH Rx,=H'xxxx' instruction which uses a literal value of 4095 and which can be easily patched in the literal pool to contain 4096. Why is this high risk behavior? I assumed, but did not state, that the person doing the patching had already checked the cross-reference list to see if the half word in the literal pool was used anywhere else where such a patch would have an adverse affect. TDIITD [1]. Please explain the details of why this is high risk behavior. Bill Fairchild [1] The Devil Is In The Details. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
