qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: QEMU for Qualcomm Hexagon - KVM Forum talk and code available


From: Taylor Simpson
Subject: RE: QEMU for Qualcomm Hexagon - KVM Forum talk and code available
Date: Tue, 17 Dec 2019 18:14:47 +0000

Hi again,

Per Aleksandar's feedback, I'm finishing up the rewrite of the code imported 
from the existing Hexagon simulator.  Once I'm finished with that work, I'll 
start breaking down the code into a patch series.  I have a couple of questions 
I hope someone would be kind enough to answer.

Question 1:
I see this error from checkpatch.pl
ERROR: Macros with complex values should be enclosed in parenthesis
However, there are times when the code will not compile with parenthesis.  For 
example, we have a file that defined all the instruction attributes.  Each line 
has
DEF_ATTRIB(LOAD, "Loads from memory", "", "")
So, we create an enum of all the possible attributes as follows
enum {
#define DEF_ATTRIB(NAME, ...) A_##NAME,
#include "attribs_def.h"
#undef DEF_ATTRIB
};
This is one simple example.  We also things like create an array of values or a 
series of small functions or cases in a switch statement.

Question 2:
What is the best source of guidance on breaking down support for a new target 
into a patch series?  I see avr being reviewed currently.  I have mostly new 
files: 12 in linux-user/hexagon, and ~50 in target/hexagon.  I also need to add 
test cases and a container for the toolchain.  Is it OK to break things down 
mostly at file boundaries?  In some cases small files could be combined, and 
large files would be broken down into reviewable chunks.  Late in the series, I 
can introduce the changes to common code.  The implication of having the common 
code changes late is that code would compile and run only at the end of the 
series.

Thanks,
Taylor


-----Original Message-----
From: Qemu-devel <qemu-devel-bounces+tsimpson=address@hidden> On Behalf Of 
Taylor Simpson
Sent: Thursday, November 14, 2019 6:54 PM
To: Richard Henderson <address@hidden>; Alex Bennée <address@hidden>; 
address@hidden
Cc: Alessandro Di Federico <address@hidden>; address@hidden; Niccolò Izzo 
<address@hidden>; Aleksandar Markovic <address@hidden>
Subject: RE: QEMU for Qualcomm Hexagon - KVM Forum talk and code available

OK, I changed the code in github and will upstream it that way.



-----Original Message-----
From: Richard Henderson <address@hidden>
Sent: Wednesday, November 13, 2019 3:10 PM
To: Taylor Simpson <address@hidden>; Alex Bennée <address@hidden>; 
address@hidden
Cc: Alessandro Di Federico <address@hidden>; address@hidden; Niccolò Izzo 
<address@hidden>; Aleksandar Markovic <address@hidden>
Subject: Re: QEMU for Qualcomm Hexagon - KVM Forum talk and code available

On 11/13/19 8:31 PM, Taylor Simpson wrote:
> [Taylor] Currently, I have the generator and the generated code sitting in 
> the source tree.  I'm flexible on this if the decision is to regenerate it 
> every time.

I would prefer to regenerate every time, and not store the generated code in 
the source tree at all.  It makes it a no-brainer to modify the source and not 
have to remember how to regenerate, because the rules are right there in the 
makefile.


r~

reply via email to

[Prev in Thread] Current Thread [Next in Thread]