| Message ID | 20150721.143904.2006742147973414532.davem@davemloft.net |
|---|---|
| State | New |
| Headers | show |
From: David Miller <davem@davemloft.net> Date: Tue, 21 Jul 2015 14:39:04 -0700 (PDT) > As currently coded, the strncmp tests dereference unaligned pointers > when setting up the test patterns. This results in SIGBUS errors > on Sparc, for example. > > Force the alignment arguments to be aligned properly for the character > type being tested. > > Ok to commit? Ping? > 2015-07-21 David S. Miller <davem@davemloft.net> > > * string/test-strncmp.c (do_test_limit): Align align1 and align2 > to CHARBYTES. > (do_test): Likewise. > > diff --git a/string/test-strncmp.c b/string/test-strncmp.c > index fb57a9b..277eb2c 100644 > --- a/string/test-strncmp.c > +++ b/string/test-strncmp.c > @@ -156,6 +156,9 @@ do_test_limit (size_t align1, size_t align2, size_t len, size_t n, int max_char, > size_t i, align_n; > CHAR *s1, *s2; > > + align1 &= ~(CHARBYTES - 1); > + align2 &= ~(CHARBYTES - 1); > + > if (n == 0) > { > s1 = (CHAR *) (buf1 + page_size); > @@ -204,6 +207,9 @@ do_test (size_t align1, size_t align2, size_t len, size_t n, int max_char, > size_t i; > CHAR *s1, *s2; > > + align1 &= ~(CHARBYTES - 1); > + align2 &= ~(CHARBYTES - 1); > + > if (n == 0) > return; >
On 07/21/2015 05:39 PM, David Miller wrote: > As currently coded, the strncmp tests dereference unaligned pointers > when setting up the test patterns. This results in SIGBUS errors > on Sparc, for example. > > Force the alignment arguments to be aligned properly for the character > type being tested. The test is designed to show that alignment of the strings should not impact the result of the testing. However, I agree that the fundamental alignment of wchar_t and char need to be preserved. The latter is easily preserved, not so with the former. > Ok to commit? > > 2015-07-21 David S. Miller <davem@davemloft.net> > > * string/test-strncmp.c (do_test_limit): Align align1 and align2 > to CHARBYTES. > (do_test): Likewise. > > diff --git a/string/test-strncmp.c b/string/test-strncmp.c > index fb57a9b..277eb2c 100644 > --- a/string/test-strncmp.c > +++ b/string/test-strncmp.c > @@ -156,6 +156,9 @@ do_test_limit (size_t align1, size_t align2, size_t len, size_t n, int max_char, > size_t i, align_n; > CHAR *s1, *s2; > > + align1 &= ~(CHARBYTES - 1); > + align2 &= ~(CHARBYTES - 1); This would make testing any align1 or align2 below CHARBYTES moot for wchar_t, when what we really want is alignment in *addition* to the minimum basic alignment for the type. Can't we shift align1 and align2 by CHARBYTES - 1 to achieve that and apply it? If you have to align please use #include <libc-internal.h> PTR_ALIGN_* or ALIGN_* macros. > + > if (n == 0) > { > s1 = (CHAR *) (buf1 + page_size); > @@ -204,6 +207,9 @@ do_test (size_t align1, size_t align2, size_t len, size_t n, int max_char, > size_t i; > CHAR *s1, *s2; > > + align1 &= ~(CHARBYTES - 1); > + align2 &= ~(CHARBYTES - 1); > + > if (n == 0) > return; > > Post v2 please. Cheers, Carlos.
diff --git a/string/test-strncmp.c b/string/test-strncmp.c index fb57a9b..277eb2c 100644 --- a/string/test-strncmp.c +++ b/string/test-strncmp.c @@ -156,6 +156,9 @@ do_test_limit (size_t align1, size_t align2, size_t len, size_t n, int max_char, size_t i, align_n; CHAR *s1, *s2; + align1 &= ~(CHARBYTES - 1); + align2 &= ~(CHARBYTES - 1); + if (n == 0) { s1 = (CHAR *) (buf1 + page_size); @@ -204,6 +207,9 @@ do_test (size_t align1, size_t align2, size_t len, size_t n, int max_char, size_t i; CHAR *s1, *s2; + align1 &= ~(CHARBYTES - 1); + align2 &= ~(CHARBYTES - 1); + if (n == 0) return;