On Sun, 12 Dec 2010, John David Anglin wrote:
> On Sun, 12 Dec 2010, Guy Martin wrote:
>
> > On Sun, 12 Dec 2010 14:21:47 -0500 (EST)
> > "John David Anglin" <dave@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > > Hmmm, I didn't think we allowed call instructions in the delay slot of
> > > a branch. Even though nullified, I think there were issues on some
> > > processors. Is an asm involved?
> >
> > I forgot to say that I've been testing this on my C3600.
> > What do you mean by "Is an asm involved?" ?
>
> Internally, the PA backend keeps track of "insn" types using an attribute
> mechanism. This is what keeps calls out of the delay slot of branches, etc.
>
> asms can do stuff that GCC doesn't know about and this can cause
> wrong code issues. I was just speculating.
This is a GCC PA bug. The code generation is wrong for this RTL:
(jump_insn 13 10 14 (set (pc)
(if_then_else (eq (zero_extract:SI (reg:SI 28 %r28 [orig:100 D.22506 ] [100])
(const_int 1 [0x1])
(const_int 28 [0x1c]))
(const_int 0 [0]))
(label_ref 16)
(pc))) include/linux/page-flags.h:297 29 {*pa.md:1610}
(expr_list:REG_BR_PRED (const_int 5 [0x5])
(expr_list:REG_DEAD (reg:SI 28 %r28 [orig:100 D.22506 ] [100])
(expr_list:REG_BR_PROB (const_int 5000 [0x1388])
(nil))))
-> 16)
(note 14 13 15 [bb 3] NOTE_INSN_BASIC_BLOCK)
(insn:TI 15 14 16 (parallel [
(asm_operands/v ("") ("") 0 []
[]
[] mm/swap_state.c:185)
(clobber (mem:BLK (scratch) [0 A8]))
]) include/linux/page-flags.h:298 -1
(nil))
(code_label 16 15 17 15 "" [1 uses])
(note 17 16 19 [bb 4] NOTE_INSN_BASIC_BLOCK)
(note 19 17 107 NOTE_INSN_DELETED)
(insn 107 19 122 (sequence [
(call_insn:TI 18 19 4 (parallel [
(set (reg:SI 28 %r28)
(call (mem:SI (symbol_ref/v:SI ("@get_swap_page") [flags 0x41] <function_decl 0x40ae6280 get_swap_page>) [0 S4 A32])
...
The output_bb code used for jump_insn 13 is confused by the memory
clobber asm. There should be a nop after the bb instruction as a
conditional branch to the following instruction is asking for disaster.
Some situations where a nop is needed are already covered, but you
have found a new one.
I created a GCC PR:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46915
This may affect other conditional branches...
Dave
--
J. David Anglin dave.anglin@xxxxxxxxxxxxxx
National Research Council of Canada (613) 990-0752 (FAX: 952-6602)
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
