aboutsummaryrefslogtreecommitdiff
path: root/gdb/printcmd.c
diff options
context:
space:
mode:
authorAndrew Burgess <andrew.burgess@embecosm.com>2023-02-10 23:49:19 +0000
committerMaciej W. Rozycki <macro@embecosm.com>2023-02-10 23:49:19 +0000
commita0c07915778486a950952139d27c01d4285b02b4 (patch)
tree4dec76630c699133cf45932f1aa6781bc7255a75 /gdb/printcmd.c
parenta2fb245a4b81ffdc93a9c6e9ceddbfb323ac9bec (diff)
downloadgdb-a0c07915778486a950952139d27c01d4285b02b4.zip
gdb-a0c07915778486a950952139d27c01d4285b02b4.tar.gz
gdb-a0c07915778486a950952139d27c01d4285b02b4.tar.bz2
GDB: Introduce limited array lengths while printing values
This commit introduces the idea of loading only part of an array in order to print it, what I call "limited length" arrays. The motivation behind this work is to make it possible to print slices of very large arrays, where very large means bigger than `max-value-size'. Consider this GDB session with the current GDB: (gdb) set max-value-size 100 (gdb) p large_1d_array value requires 400 bytes, which is more than max-value-size (gdb) p -elements 10 -- large_1d_array value requires 400 bytes, which is more than max-value-size notice that the request to print 10 elements still fails, even though 10 elements should be less than the max-value-size. With a patched version of GDB: (gdb) p -elements 10 -- large_1d_array $1 = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9...} So now the print has succeeded. It also has loaded `max-value-size' worth of data into value history, so the recorded value can be accessed consistently: (gdb) p -elements 10 -- $1 $2 = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9...} (gdb) p $1 $3 = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, <unavailable> <repeats 75 times>} (gdb) Accesses with other languages work similarly, although for Ada only C-style [] array element/dimension accesses use history. For both Ada and Fortran () array element/dimension accesses go straight to the inferior, bypassing the value history just as with C pointers. Co-Authored-By: Maciej W. Rozycki <macro@embecosm.com>
Diffstat (limited to 'gdb/printcmd.c')
-rw-r--r--gdb/printcmd.c16
1 files changed, 16 insertions, 0 deletions
diff --git a/gdb/printcmd.c b/gdb/printcmd.c
index 7f35513..9bda60d 100644
--- a/gdb/printcmd.c
+++ b/gdb/printcmd.c
@@ -1242,6 +1242,11 @@ print_command_parse_format (const char **expp, const char *cmdname,
void
print_value (value *val, const value_print_options &opts)
{
+ /* This setting allows large arrays to be printed by limiting the
+ number of elements that are loaded into GDB's memory; we only
+ need to load as many array elements as we plan to print. */
+ scoped_array_length_limiting limit_large_arrays (opts.print_max);
+
int histindex = record_latest_value (val);
annotate_value_history_begin (histindex, value_type (val));
@@ -1301,6 +1306,11 @@ process_print_command_args (const char *args, value_print_options *print_opts,
if (exp != nullptr && *exp)
{
+ /* This setting allows large arrays to be printed by limiting the
+ number of elements that are loaded into GDB's memory; we only
+ need to load as many array elements as we plan to print. */
+ scoped_array_length_limiting limit_large_arrays (print_opts->print_max);
+
/* VOIDPRINT is true to indicate that we do want to print a void
value, so invert it for parse_expression. */
expression_up expr = parse_expression (exp, nullptr, !voidprint);
@@ -1489,6 +1499,12 @@ output_command (const char *exp, int from_tty)
get_formatted_print_options (&opts, format);
opts.raw = fmt.raw;
+
+ /* This setting allows large arrays to be printed by limiting the
+ number of elements that are loaded into GDB's memory; we only
+ need to load as many array elements as we plan to print. */
+ scoped_array_length_limiting limit_large_arrays (opts.print_max);
+
print_formatted (val, fmt.size, &opts, gdb_stdout);
annotate_value_end ();