3 * Since using watchpoints can be very slow, we have to take some pains to
4 * ensure that we don't run too long with them enabled or we run the risk
5 * of having the test timeout. To help avoid this, we insert some marker
6 * functions in the execution stream so we can set breakpoints at known
7 * locations, without worrying about invalidating line numbers by changing
8 * this file. We use null bodied functions are markers since gdb does
9 * not support breakpoints at labeled text points at this time.
11 * One place we need is a marker for when we start executing our tests
12 * instructions rather than any process startup code, so we insert one
13 * right after entering main(). Another is right before we finish, before
14 * we start executing any process termination code.
16 * Another problem we have to guard against, at least for the test
17 * suite, is that we need to ensure that the line that causes the
18 * watchpoint to be hit is still the current line when gdb notices
19 * the hit. Depending upon the specific code generated by the compiler,
20 * the instruction after the one that triggers the hit may be part of
21 * the same line or part of the next line. Thus we ensure that there
22 * are always some instructions to execute on the same line after the
23 * code that should trigger the hit.
36 struct foo struct1
, struct2
, *ptr1
, *ptr2
;
64 /* The point of this is that we will set a breakpoint at this call.
66 Then, if DECR_PC_AFTER_BREAK equals the size of a function call
67 instruction (true on a sun3 if this is gcc-compiled--FIXME we
68 should use asm() to make it work for any compiler, present or
69 future), then we will end up branching to the location just after
70 the breakpoint. And we better not confuse that with hitting the
88 for (count
= 0; count
< 4; count
++) {
90 ival3
= count
; ival4
= count
;
92 ival1
= count
; /* Outside loop */
94 ival3
= count
; ival4
= count
;
98 static char msg
[] = "type stuff for buf now:";
99 write (1, msg
, sizeof (msg
) - 1);
100 read (0, &buf
[0], 5);
104 /* We have a watchpoint on ptr1->val. It should be triggered if
105 ptr1's value changes. */
108 /* This should not trigger the watchpoint. If it does, then we
109 used the wrong value chain to re-insert the watchpoints or we
110 are not evaluating the watchpoint expression correctly. */
114 /* We have a watchpoint on ptr1->val. It should be triggered if
115 ptr1's value changes. */
118 /* This should not trigger the watchpoint. If it does, then we
119 used the wrong value chain to re-insert the watchpoints or we
120 are not evaluating the watchpoint expression correctly. */
This page took 0.039648 seconds and 4 git commands to generate.