4187119d |
1 | |
2 | @node Convex,,, Top |
3 | @appendix Convex-specific info |
4 | @cindex Convex notes |
5 | |
6 | Scalar registers are 64 bits long, which is a pain since |
7 | left half of an S register frequently contains noise. |
8 | Therefore there are two ways to obtain the value of an S register. |
9 | |
10 | @table @kbd |
11 | @item $s0 |
12 | returns the low half of the register as an int |
13 | |
14 | @item $S0 |
15 | returns the whole register as a long long |
16 | @end table |
17 | |
18 | You can print the value in floating point by using @samp{p/f $s0} or @samp{p/f $S0} |
19 | to print a single or double precision value. |
20 | |
21 | @cindex vector registers |
22 | Vector registers are handled similarly, with @samp{$V0} denoting the whole |
23 | 64-bit register and @kbd{$v0} denoting the 32-bit low half; @samp{p/f $v0} |
24 | or @samp{p/f $V0} can be used to examine the register in floating point. |
25 | The length of the vector registers is taken from @samp{$vl}. |
26 | |
27 | Individual elements of a vector register are denoted in the obvious way; |
28 | @samp{print $v3[9]} prints the tenth element of register @kbd{v3}, and |
29 | @samp{set $v3[9] = 1234} alters it. |
30 | |
31 | @kbd{$vl} and @kbd{$vs} are int, and @kbd{$vm} is an int vector. |
32 | Elements of @kbd{$vm} can't be assigned to. |
33 | |
34 | @cindex communication registers |
35 | @kindex info comm-registers |
36 | Communication registers have names @kbd{$C0 .. $C63}, with @kbd{$c0 .. $c63} |
37 | denoting the low-order halves. @samp{info comm-registers} will print them |
38 | all out, and tell which are locked. (A communication register is |
39 | locked when a value is sent to it, and unlocked when the value is |
40 | received.) Communication registers are, of course, global to all |
41 | threads, so it does not matter what the currently selected thread is. |
42 | @samp{info comm-reg @var{name}} prints just that one communication |
43 | register; @samp{name} may also be a communication register number |
44 | @samp{nn} or @samp{0xnn}. |
45 | @samp{info comm-reg @var{address}} prints the contents of the resource |
46 | structure at that address. |
47 | |
48 | @kindex info psw |
49 | The command @samp{info psw} prints the processor status word @kbd{$ps} |
50 | bit by bit. |
51 | |
52 | @kindex set base |
53 | GDB normally prints all integers in base 10, but the leading |
54 | @kbd{0x80000000} of pointers is intolerable in decimal, so the default |
55 | output radix has been changed to try to print addresses appropriately. |
56 | The @samp{set base} command can be used to change this. |
57 | |
58 | @table @code |
59 | @item set base 10 |
60 | Integer values always print in decimal. |
61 | |
62 | @item set base 16 |
63 | Integer values always print in hex. |
64 | |
65 | @item set base |
66 | Go back to the initial state, which prints integer values in hex if they |
67 | look like pointers (specifically, if they start with 0x8 or 0xf in the |
68 | stack), otherwise in decimal. |
69 | @end table |
70 | |
71 | @kindex set pipeline |
72 | When an exception such as a bus error or overflow happens, usually the PC |
73 | is several instructions ahead by the time the exception is detected. |
74 | The @samp{set pipe} command will disable this. |
75 | |
76 | @table @code |
77 | @item set pipeline off |
78 | Forces serial execution of instructions; no vector chaining and no |
79 | scalar instruction overlap. With this, exceptions are detected with |
80 | the PC pointing to the instruction after the one in error. |
81 | |
82 | @item set pipeline on |
83 | Returns to normal, fast, execution. This is the default. |
84 | @end table |
85 | |
86 | @cindex parallel |
87 | In a parallel program, multiple threads may be executing, each |
88 | with its own registers, stack, and local memory. When one of them |
89 | hits a breakpoint, that thread is selected. Other threads do |
90 | not run while the thread is in the breakpoint. |
91 | |
92 | @kindex 1cont |
93 | The selected thread can be single-stepped, given signals, and so |
94 | on. Any other threads remain stopped. When a @samp{cont} command is given, |
95 | all threads are resumed. To resume just the selected thread, use |
96 | the command @samp{1cont}. |
97 | |
98 | @kindex thread |
99 | The @samp{thread} command will show the active threads and the |
100 | instruction they are about to execute. The selected thread is marked |
101 | with an asterisk. The command @samp{thread @var{n}} will select thread @var{n}, |
102 | shifting the debugger's attention to it for single-stepping, |
103 | registers, local memory, and so on. |
104 | |
105 | @kindex info threads |
106 | The @samp{info threads} command will show what threads, if any, have |
107 | invisibly hit breakpoints or signals and are waiting to be noticed. |
108 | |
109 | @kindex set parallel |
110 | The @samp{set parallel} command controls how many threads can be active. |
111 | |
112 | @table @code |
113 | @item set parallel off |
114 | One thread. Requests by the program that other threads join in |
115 | (spawn and pfork instructions) do not cause other threads to start up. |
116 | This does the same thing as the @samp{limit concurrency 1} command. |
117 | |
118 | @item set parallel fixed |
119 | All CPUs are assigned to your program whenever it runs. When it |
120 | executes a pfork or spawn instruction, it begins parallel execution |
121 | immediately. This does the same thing as the @samp{mpa -f} command. |
122 | |
123 | @item set parallel on |
124 | One or more threads. Spawn and pfork cause CPUs to join in when and if |
125 | they are free. This is the default. It is very good for system |
126 | throughput, but not very good for finding bugs in parallel code. If you |
127 | suspect a bug in parallel code, you probably want @samp{set parallel fixed.} |
128 | @end table |
129 | |
130 | @subsection Limitations |
131 | |
132 | WARNING: Convex GDB evaluates expressions in long long, because S |
133 | registers are 64 bits long. However, GDB expression semantics are not |
134 | exactly C semantics. This is a bug, strictly speaking, but it's not one I |
135 | know how to fix. If @samp{x} is a program variable of type int, then it |
136 | is also type int to GDB, but @samp{x + 1} is long long, as is @samp{x + y} |
137 | or any other expression requiring computation. So is the expression |
138 | @samp{1}, or any other constant. You only really have to watch out for |
139 | calls. The innocuous expression @samp{list_node (0x80001234)} has an |
140 | argument of type long long. You must explicitly cast it to int. |
141 | |
142 | It is not possible to continue after an uncaught fatal signal by using |
143 | @samp{signal 0}, @samp{return}, @samp{jump}, or anything else. The difficulty is with |
144 | Unix, not GDB. |
145 | |
146 | I have made no big effort to make such things as single-stepping a |
147 | @kbd{join} instruction do something reasonable. If the program seems to |
148 | hang when doing this, type @kbd{ctrl-c} and @samp{cont}, or use |
149 | @samp{thread} to shift to a live thread. Single-stepping a @kbd{spawn} |
150 | instruction apparently causes new threads to be born with their T bit set; |
151 | this is not handled gracefully. When a thread has hit a breakpoint, other |
152 | threads may have invisibly hit the breakpoint in the background; if you |
153 | clear the breakpoint gdb will be surprised when threads seem to continue |
154 | to stop at it. All of these situations produce spurious signal 5 traps; |
155 | if this happens, just type @samp{cont}. If it becomes a nuisance, use |
156 | @samp{handle 5 nostop}. (It will ask if you are sure. You are.) |
157 | |
158 | There is no way in GDB to store a float in a register, as with |
159 | @kbd{set $s0 = 3.1416}. The identifier @kbd{$s0} denotes an integer, |
160 | and like any C expression which assigns to an integer variable, the |
161 | right-hand side is casted to type int. If you should need to do |
162 | something like this, you can assign the value to @kbd{@{float@} ($sp-4)} |
163 | and then do @kbd{set $s0 = $sp[-4]}. Same deal with @kbd{set $v0[69] = 6.9}. |