<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Tahoma
}
--></style>
</head>
<body class='hmmessage'>
Great answer, now it works. Only, I now have the immediately following problem:<br><br>I insert the copies from the 32 bit load, like<br><br> %reg16507<def> = ld %r4<br> %reg16457<def> = COPY %reg16507:lo16<br> %reg16443<def> = COPY %reg16507:hi16<br> %reg16510<def> = ld %r5<br> %reg16458<def> = COPY %reg16510:lo16<br> %reg16444<def> = COPY %reg16510:hi16<br> %reg16468<def> = add <u>%reg16457</u>, %reg16458 ; 16457 is regclass high-part <br><br>I make two loads, and copy high and low parts respectively to 16 bit registers for use, as you recommended.<br>The final instruction in the list is an addition, which is however onstrained to use the high subregister (%16457). The low subregister is copied to a virtual register with the hi16-regclass, but this gets coalesced to the wrong regclass.<br><br>The problem is that when I follow the comment for getMatchingSuperRegClass() - as wittingly I can - I then simply return A, as it is a proper register class<br>containing all registers in B; B is the regclass containing all subregisters of A with only the high parts. So, if A has Reg32_1, B has Reg16_1_hi, and so on.<br><br>So, when I return A, the COPY gets coalesced, but the registerclass for the new interval becomes the one of 32 bits, which gives an error for the addition instruction.<br><br>I would like to ask for help in as to what this method should actually do - if I am missing it, or if not, I then wonder what register classes I should use to make this work?<br><br>I perceive this: A={32 bit regs}, B={high parts of the registers in A}, so if called with (A,B,:hi), return A.<br><br>Is there something else missing here?<br><br>Thanks,<br><br>Jonas<br><br>> Subject: Re: [LLVMdev] wide memory accesses<br>> From: stoklund@2pi.dk<br>> Date: Mon, 9 May 2011 09:51:37 -0700<br>> CC: llvmdev@cs.uiuc.edu<br>> To: jnspaulsson@hotmail.com<br>> <br>> <br>> On May 9, 2011, at 9:00 AM, Jonas Paulsson wrote:<br>> <br>> > Hi,<br>> > <br>> > I am trying to take 16 bit memory reads and combine them to a single 32 bit read. I am having trouble to make the code simply read 32 bytes and the use the subregisters accordingly, without unnecessary copying.<br>> > <br>> > I have tried two techniques, in the MachineFunction:<br>> > <br>> > 1. replace the MachineOperands in the users of the data with the new register/subregister index. This yields an assert failure during VirtRegRewriter, in substPhysReg: "Invalid SubReg for physical register", after the Two-address rewrote this:<br>> > <br>> > %reg16445<def> = add %reg16507:hi16, %reg16510:hi16 ; 32bit:16507,16510, 16bit: 16445 <br>> > prepend: %reg16445<def> = COPY %reg16507; <br>> > rewrite to: %reg16445<def> = addh_1_8 %reg16445:hi16, %reg16510:hi16<br>> > <br>> > In my eyes, there should not have been a subreg 'hi16' for the 16445 reg - this reg is 16 bits. I would have wished that the 16507:hi16 be interpreted as the corresponding subregister, and thus generated in the COPY with a :hi16. It is all right that the 16445 is of 16 bits, this is correct, but then it is used incorrectly with a :hi16 subregister value ?? Any ideas?<br>> <br>> If you are doing this before the register allocator passes, you must make sure that the code preserves SSA form. That can be difficult to do when dealing with sub-registers. Other targets just emit EXTRACT_SUBREG / INSERT_SUBREG and let the coalescer deal with it.<br>> <br>> It looks like TwoAddressInctructionPass is not ready to deal with subreg indexes either, it is creating wrong code in your example.<br>> <br>> I recommend:<br>> <br>> > 2. Insert COPY's, but these would not get coalesced away, so instead of saving instructions I ended up with one load and two moves...:-( How could I get the wide load to simply be used intelligently by COPYing to the old virtual registers?<br>> <br>> You need to implement getMatchingSuperRegClass in your register info class. That will give the coalescer the needed information to join subreg copies.<br>> <br>> /jakob<br>> <br> </body>
</html>