hw/arm/smmuv3-internal.h: Don't use locals in statement macros

The STE_CTXPTR() and STE_S2TTB() macros both extract two halves
of an address from fields in the STE and combine them into a
single value to return. The current code for this uses a GCC
statement expression. There are two problems with this:

(1) The type chosen for the variable in the statement expr
is 'unsigned long', which might not be 64 bits

(2) the name chosen for the variable causes -Wshadow warnings
because it's the same as a variable in use at the callsite:

In file included from ../../hw/arm/smmuv3.c:34:
../../hw/arm/smmuv3.c: In function ‘smmu_get_cd’:
../../hw/arm/smmuv3-internal.h:538:23: warning: declaration of ‘addr’ shadows a previous local [-Wshadow=compatible-local]
  538 |         unsigned long addr;                                     \
      |                       ^~~~
../../hw/arm/smmuv3.c:339:23: note: in expansion of macro ‘STE_CTXPTR’
  339 |     dma_addr_t addr = STE_CTXPTR(ste);
      |                       ^~~~~~~~~~
../../hw/arm/smmuv3.c:339:16: note: shadowed declaration is here
  339 |     dma_addr_t addr = STE_CTXPTR(ste);
      |                ^~~~

Sidestep both of these problems by just using a single
expression rather than a statement expr.

For CMD_ADDR, we got the type of the variable right but still
run into -Wshadow problems:

In file included from ../../hw/arm/smmuv3.c:34:
../../hw/arm/smmuv3.c: In function ‘smmuv3_range_inval’:
../../hw/arm/smmuv3-internal.h:334:22: warning: declaration of ‘addr’ shadows a previous local [-Wshadow=compatible-local]
  334 |             uint64_t addr = high << 32 | (low << 12);         \
      |                      ^~~~
../../hw/arm/smmuv3.c:1104:28: note: in expansion of macro ‘CMD_ADDR’
 1104 |     dma_addr_t end, addr = CMD_ADDR(cmd);
      |                            ^~~~~~~~
../../hw/arm/smmuv3.c:1104:21: note: shadowed declaration is here
 1104 |     dma_addr_t end, addr = CMD_ADDR(cmd);
      |                     ^~~~

so convert it too.

CD_TTB has neither problem, but it is the only other macro in
the file that uses this pattern, so we convert it also for
consistency's sake.

We use extract64() rather than extract32() to avoid having
to explicitly cast the result to uint64_t.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-ID: <20230922152944.3583438-5-peter.maydell@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

0 files changed, 0 insertions, 0 deletions