The attached patch corrects a couple of typos in argument numbers
in the handling of __builtin__vsnprintf_chk calls in the gimple-
ssa-sprintf pass, and another couple of typos in the test for
this that were masking this failure.
As an aside, the patch also fixes the off-by-one line test failures
introduced in r240503. If there is a way to make the line numbers
relative (as suggested in
https://gcc.gnu.org/ml/gcc-patches/2016-09/msg02070.html) I'm happy
to update the -Wformat-length tests to make use of them (and document
it on the Wiki) if someone can point me at an example (or
documentation). I couldn't find examples of dg-warning directives
that use the feature.
Thanks
Martin
PR c/77762 - Incorrect destination buffer length in -Wformat-length warning
gcc/testsuite/ChangeLog:
2016-09-27 Martin Sebor <mse...@redhat.com>
PR c/77762
* gcc.dg/tree-ssa/builtin-sprintf-warn-1.c (test_vsnprintf_chk_s):
Call __builtin___vsnprintf_chk, not __builtin___snprintf_chk.
(test_sprintf_p_const): Adjust line numbers to avoid failures
introduced in r240503.
gcc/ChangeLog:
2016-09-27 Martin Sebor <mse...@redhat.com>
PR c/77762
* gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
Fix typos.
Index: gcc/gimple-ssa-sprintf.c
===================================================================
--- gcc/gimple-ssa-sprintf.c (revision 240556)
+++ gcc/gimple-ssa-sprintf.c (working copy)
@@ -2536,8 +2536,8 @@ pass_sprintf_length::handle_gimple_call (gimple_st
// Signature:
// __builtin___vsnprintf_chk (dst, size, ost, objsize, format, va)
idx_dstsize = 1;
- idx_objsize = 2;
- idx_format = 3;
+ idx_objsize = 3;
+ idx_format = 4;
info.argidx = -1;
info.bounded = true;
break;
Index: gcc/testsuite/gcc.dg/tree-ssa/builtin-sprintf-warn-1.c
===================================================================
--- gcc/testsuite/gcc.dg/tree-ssa/builtin-sprintf-warn-1.c (revision 240556)
+++ gcc/testsuite/gcc.dg/tree-ssa/builtin-sprintf-warn-1.c (working copy)
@@ -95,7 +95,7 @@ void test_sprintf_p_const (void)
format null pointers as 0 or 0x0 and so the following will only be
diagnosed on the former targets. */
T (5, "%p", (void*)0);
- /* { dg-warning "nul past the end" "(nil)" { target *-linux-gnu *-*-uclinux } 96 } */
+ /* { dg-warning "nul past the end" "(nil)" { target *-linux-gnu *-*-uclinux } 97 } */
/* The exact output for %p is unspecified by C. Two formats are known:
same as %tx (for example AIX) and same as %#tx (for example Solaris). */
@@ -107,8 +107,8 @@ void test_sprintf_p_const (void)
as with signed integer conversions (i.e., it prepends a space). Other
known implementations ignore it. */
T (6, "% p", (void*)0x234); /* { dg-warning ". . flag used with .%p." } */
- /* { dg-warning "nul past the end" "Glibc %p" { target *-linux-gnu } 108 } */
- /* { dg-warning "nul past the end" "Generic %p" { target *-*-uclinux } 108 } */
+ /* { dg-warning "nul past the end" "Glibc %p" { target *-linux-gnu } 109 } */
+ /* { dg-warning "nul past the end" "Generic %p" { target *-*-uclinux } 109 } */
}
/* Verify that no warning is issued for calls that write into a flexible
@@ -1404,9 +1404,9 @@ void test_vsnprintf_chk_s (__builtin_va_list va)
/* Verify that specifying a size of the destination buffer that's
bigger than its actual size (normally determined and passed to
the function by __builtin_object_size) is diagnosed. */
- __builtin___snprintf_chk (buffer, 123, 0, 122, " "); /* { dg-warning "always overflow|specified size 123 exceeds the size 122 of the destination object" } */
+ __builtin___vsnprintf_chk (buffer, 123, 0, 122, "%-s", va); /* { dg-warning "always overflow|specified size 123 exceeds the size 122 of the destination object" } */
- __builtin___snprintf_chk (buffer, __SIZE_MAX__, 0, 2, " "); /* { dg-warning "always overflow|destination size .\[0-9\]+. too large" } */
+ __builtin___vsnprintf_chk (buffer, __SIZE_MAX__, 0, 2, "%-s", va); /* { dg-warning "always overflow|destination size .\[0-9\]+. too large" } */
T (0, "%s");
T (1, "%s");
