Fix amd64-gcc build

[lwhsu]

Summary

Fix amd64-gcc build by removing declarations already in amd64/include/counter.h

Error messages:

/workspace/src/sys/amd64/amd64/pmap.c:446:21: error: redundant redeclaration of '__pcpu' [-Werror=redundant-decls]
 extern struct pcpu *__pcpu;
                     ^~~~~~
In file included from /workspace/src/sys/sys/counter.h:37:0,
                 from /workspace/src/sys/sys/vmmeter.h:65,
                 from /workspace/src/sys/amd64/amd64/pmap.c:130:
./machine/counter.h:36:21: note: previous declaration of '__pcpu' was here
 extern struct pcpu *__pcpu;

/workspace/src/sys/amd64/amd64/pmap.c:447:20: error: redundant redeclaration of 'temp_bsp_pcpu' [-Werror=redundant-decls]
 extern struct pcpu temp_bsp_pcpu;
                    ^~~~~~~~~~~~~
In file included from /workspace/src/sys/sys/counter.h:37:0,
                 from /workspace/src/sys/sys/vmmeter.h:65,
                 from /workspace/src/sys/amd64/amd64/pmap.c:130:
./machine/counter.h:37:20: note: previous declaration of 'temp_bsp_pcpu' was here
 extern struct pcpu temp_bsp_pcpu;
                    ^~~~~~~~~~~~~

Test Plan

make world and kernel with CROSS_TOOLCHAIN=amd64-gcc

[markj]

sys/amd64/amd64/pmap.c

446

This line can be deleted too, it shadows a local var in pmap_bootstrap().

[kib]

I disagree with making pmap/pcpu define live in counter.h.

[lwhsu]

In D21417#466003, @kib wrote:

I disagree with making pmap/pcpu define live in counter.h.

I also feel these in counter.h is a bit odd, but we have extern struct pcpu __pcpu[]; in all sys/<arch>/include/counter.h

Is the right solution is removing all these definitions from all counter.h?

[kib]

In D21417#466004, @lwhsu wrote:

In D21417#466003, @kib wrote:

I disagree with making pmap/pcpu define live in counter.h.

I also feel these in counter.h is a bit odd, but we have extern struct pcpu __pcpu[]; in all sys/<arch>/include/counter.h

Is the right solution is removing all these definitions from all counter.h?

If it works, but I am sure it does not.

I have some plans about this stuff, I probably post something later today.

[lwhsu]

Suppressed by D21418

[lwhsu]

@kib : I found that this one and D21418 do not conflict, do you mind if I commit this first to fix gcc build?

[kib]

In D21417#466378, @lwhsu wrote:

@kib : I found that this one and D21418 do not conflict, do you mind if I commit this first to fix gcc build?

Please do not.
I think you should have (or provide) a way to specify -Wno-error for emergency while the issues are sorted, for non-standard toolchain builds.

[lwhsu]

Wait for D21418.

sys/amd64/amd64/pmap.c

#define td_retval       td_uretoff.tdu_retval
         struct td_sched        *td_sched;        /* (*) Scheduler-specific data. */
         struct kaudit_record        *td_ar;        /* (k) Active audit record, if any. */
         struct lpohead        td_lprof[2];        /* (a) lock profiling objects. */
         struct kdtrace_thread        *td_dtrace; /* (*) DTrace-specific data. */
         int             td_errno;       /* Error returned by last syscall. */
         struct vnet        *td_vnet;        /* (k) Effective vnet. */
         const char        *td_vnet_lpush;        /* (k) Debugging vnet push / pop. */
         struct trapframe *td_intr_frame;/* (k) Frame of the current irq */
-        struct proc        *td_rfppwait_p;        /* (k) The vforked child */
+        struct thread        *td_forked_td;        /* (k) The forked child */
         struct vm_page        **td_ma;        /* (k) uio pages held */
         int             td_ma_cnt;      /* (k) size of *td_ma */
         void            *td_emuldata;   /* Emulator state data */
 };
 
 struct mtx *thread_lock_block(struct thread *);
 void thread_lock_unblock(struct thread *, struct mtx *);
 void thread_lock_set(struct thread *, struct mtx *);