[patch][rfa]@arch@_cgen_ifld_table

[Dave Brolley]

Hi,

Stan Cox noticed the following when debugging a problem with one of our 
internal ports. I've replaced the name of the port with '@arch@' and 
'@ARCH@' below:

-------------------------------------------------------------------------

I notice that @arch@-opc.c defines:
 #define F(f) & @arch@_cgen_ifld_table[@ARCH@_ ##f]
so it would seem as though the members of @arch@_cgen_ifld_table should
match those of enum ifield_type.   However they do not always do so;
they seem to get out of sync for multi-ifields.

(@arch@-desc.c)
  { @ARCH@_F_CSRN_HI, "f-csrn-hi", 0, 32, 15, 1, { 0, { { { (1<<MACH_BASE), 0 } } } } },
  { @ARCH@_F_CSRN_LO, "f-csrn-lo", 0, 32, 8, 4, { 0, { { { (1<<MACH_BASE), 0 } } } }  },
  { @ARCH@_F_CRNX_HI, "f-crnx-hi", 0, 32, 28, 1, { 0, { { { (1<<MACH_BASE), 0 } } } } },
  { @ARCH@_F_CRNX_LO, "f-crnx-lo", 0, 32, 4, 4, { 0, { { { (1<<MACH_BASE), 0 } } } }  },

versus (@arch@-desc.h)

 , @ARCH@_F_CSRN_HI, @ARCH@_F_CSRN_LO, @ARCH@_F_CSRN, @ARCH@_F_CRNX_HI
 , @ARCH@_F_CRNX_LO, @ARCH@_F_CRNX, @ARCH@_F_0, @ARCH@_F_1

---------------------------------------------------------------------

As it turns out, gen-ifld-defns does explitictely omit VIRTUAL and 
derived ifields when generating @arch@_cgen_ifld_table and 
gen-maybe-multi-ifield expects them to be omitted when indexing the 
table using ifld-number which generates hard coded indices. However, 
Stan is correct that the generated macros 'F' index the table using the 
enumerators and so require that all ifields be in the table in the 
correct order.

The attached patch changes gen-maybe-multi-ifield to also index the 
table using the enumerators (much cleaner) and adds the missing ifields 
to the table. Note that some information is not available for the 
additional fields (ifld-word-offset, ifld-word-length, ifld-start, 
ifld-length) and zeroes are generated in these cases.

ifld-number is no longer used. Should it be removed?

I've also attached a diff of fr30-desc.c so you can see how the 
generated code changes.

This has been tested on fr30 and on our internal port.

Seeking approval to commit.

Dave

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ifld_table.patch.txt
URL: <http://sourceware.org/pipermail/cgen/attachments/20030211/3e7ff625/attachment.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: fr30.diff.txt
URL: <http://sourceware.org/pipermail/cgen/attachments/20030211/3e7ff625/attachment-0001.txt>