decodetree: Allow custom var width load functions

This is useful in situations where you want decodetree
to handle variable width instructions but you want to
provide custom code to load the instructions. Suppressing
the generation of the load function is necessary to avoid
compilation errors due to the load function being unused.

This will be used by the PowerPC decodetree code.

Signed-off-by: Luis Pires <luis.pires@eldorado.org.br>
---
 scripts/decodetree.py | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

Please ignore this. I'll resend as part of a patch series.

-----Original Message-----
From: Luis Pires <luis.pires@eldorado.org.br> 
Sent: terça-feira, 13 de abril de 2021 15:16
To: qemu-devel@nongnu.org
Cc: richard.henderson@linaro.org; qemu-ppc@nongnu.org; Luis Fernando Fujita Pires <luis.pires@eldorado.org.br>
Subject: [PATCH] decodetree: Allow custom var width load functions

This is useful in situations where you want decodetree to handle variable width instructions but you want to provide custom code to load the instructions. Suppressing the generation of the load function is necessary to avoid compilation errors due to the load function being unused.

This will be used by the PowerPC decodetree code.

Signed-off-by: Luis Pires <luis.pires@eldorado.org.br>
---
 scripts/decodetree.py | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/scripts/decodetree.py b/scripts/decodetree.py index fef5eeaf42..c88dbdb4c3 100644
--- a/scripts/decodetree.py
+++ b/scripts/decodetree.py
@@ -1277,9 +1277,10 @@ def main():
     global anyextern
 
     decode_scope = 'static '
+    noloadfunc = False
 
     long_opts = ['decode=', 'translate=', 'output=', 'insnwidth=',
-                 'static-decode=', 'varinsnwidth=']
+                 'static-decode=', 'varinsnwidth=', 'noloadfunc']
     try:
         (opts, args) = getopt.gnu_getopt(sys.argv[1:], 'o:vw:', long_opts)
     except getopt.GetoptError as err:
@@ -1311,6 +1312,8 @@ def main():
                 deposit_function = 'deposit64'
             elif insnwidth != 32:
                 error(0, 'cannot handle insns of width', insnwidth)
+        elif o == '--noloadfunc':
+            noloadfunc = True
         else:
             assert False, 'unhandled option'
 
@@ -1400,12 +1403,13 @@ def main():
     output(i4, 'return false;\n')
     output('}\n')
 
-    if variablewidth:
-        output('\n', decode_scope, insntype, ' ', decode_function,
-               '_load(DisasContext *ctx)\n{\n',
-               '    ', insntype, ' insn = 0;\n\n')
-        stree.output_code(4, 0, 0, 0)
-        output('}\n')
+    if not noloadfunc:
+        if variablewidth:
+            output('\n', decode_scope, insntype, ' ', decode_function,
+                   '_load(DisasContext *ctx)\n{\n',
+                   '    ', insntype, ' insn = 0;\n\n')
+            stree.output_code(4, 0, 0, 0)
+            output('}\n')
 
     if output_file:
         output_fd.close()
--
2.25.1

On 4/13/21 11:16 AM, Luis Pires wrote:
> This is useful in situations where you want decodetree
> to handle variable width instructions but you want to
> provide custom code to load the instructions. Suppressing
> the generation of the load function is necessary to avoid
> compilation errors due to the load function being unused.
> 
> This will be used by the PowerPC decodetree code.
> 
> Signed-off-by: Luis Pires<luis.pires@eldorado.org.br>
> ---
>   scripts/decodetree.py | 18 +++++++++++-------
>   1 file changed, 11 insertions(+), 7 deletions(-)

So, in the previous cases where we have very controlled "variable length", 
we've simply created separate decode files each with fixed length.

E.g. for arm thumb, in thumb_tr_translate_insn:

     insn = arm_lduw_code(env, dc->base.pc_next, dc->sctlr_b);
     is_16bit = thumb_insn_is_16bit(dc, dc->base.pc_next, insn);
...
     if (is_16bit) {
         disas_thumb_insn(dc, insn);
     } else {
         disas_thumb2_insn(dc, insn);
     }

Similarly, riscv's decode_opc.

I would think that Power would work the same, with 32-bit and 64-bit 
instructions.  I'd like you to talk me through what you want to do here.


r~