diff options
| author | Joseph Myers <joseph@codesourcery.com> | 2015-11-23 08:50:53 +0000 |
|---|---|---|
| committer | Joseph Myers <joseph@codesourcery.com> | 2015-11-23 08:50:53 +0000 |
| commit | f5eee5c72b2ab56f3faf4f46729fe82805efde68 (patch) | |
| tree | 5835c3e294687c8019b6396ffef2fc758ecf4d66 /stdlib/Makefile | |
| parent | f549f0bcba7196a2afc51657c536bbc131a7c544 (diff) | |
| download | glibc-f5eee5c72b2ab56f3faf4f46729fe82805efde68.tar.gz glibc-f5eee5c72b2ab56f3faf4f46729fe82805efde68.tar.xz glibc-f5eee5c72b2ab56f3faf4f46729fe82805efde68.zip | |
Fix strtol in Turkish locales (bug 19242).
The implementations of strtol and related functions use locale-specific conversions to upper case before determining whether a character is a valid letter in the argument. This means that in Turkish locales such as tr_TR.UTF-8 and tr_TR.ISO-8859-9, "i" is interpreted as not being a valid number, when if the base passed to strtol is 19 or more it should be interpreted as the number 18. ISO C explicitly says "The letters from a (or A) through z (or Z) are ascribed the values 10 through 35", so clearly intends the standard ASCII letters (otherwise you wouldn't generally have exactly 26 letters to ascribe such values) (whereas white-space must be identified according to the locale). In particular, 'i' and 'I' must be understood to be in that sequence. This patch makes the code do the case conversions and classification in the C locale; the user's locale remains used for whitespace testing (explicitly correct according to ISO C). Note that the way the code worked, the only non-ASCII letter that would previously have been accepted would have been the Turkish 'ı' (dotless 'i'), because the uppercase version of that in Turkish locales is 'I'. This patch means that will no longer be accepted, which seems appropriate. Tested for x86_64 and x86. [BZ #19242] * stdlib/strtol_l.c (ISALPHA): Use _nl_C_locobj_ptr for locale. (TOUPPER): Likewise. * stdlib/tst-strtol-locale-main.c: New file. * stdlib/tst-strtol-locale.c: Likewise. * stdlib/Makefile (tests): Add tst-strtol-locale. [$(run-built-tests) = yes] (LOCALES): Add tr_TR.ISO-8859-9. [$(run-built-tests) = yes] ($(objpfx)tst-strtol-locale.out): Depend on $(gen-locales). * wcsmbs/tst-wcstol-locale.c: New file. * wcsmbs/Makefile (tests): Add tst-wcstol-locale. [$(run-built-tests) = yes] (LOCALES): Add tr_TR.UTF-8 and tr_TR.ISO-8859-9. [$(run-built-tests) = yes] ($(objpfx)tst-wcstol-locale.out): Depend on $(gen-locales).
Diffstat (limited to 'stdlib/Makefile')
| -rw-r--r-- | stdlib/Makefile | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/stdlib/Makefile b/stdlib/Makefile index 9e0c2495c1..e8b5b8cc1b 100644 --- a/stdlib/Makefile +++ b/stdlib/Makefile @@ -74,7 +74,8 @@ tests := tst-strtol tst-strtod testmb testrand testsort testdiv \ tst-makecontext3 bug-getcontext bug-fmtmsg1 \ tst-secure-getenv tst-strtod-overflow tst-strtod-round \ tst-tininess tst-strtod-underflow tst-tls-atexit \ - tst-setcontext3 tst-tls-atexit-nodelete + tst-setcontext3 tst-tls-atexit-nodelete \ + tst-strtol-locale tests-static := tst-secure-getenv modules-names = tst-tls-atexit-lib @@ -122,7 +123,8 @@ endif include ../Rules ifeq ($(run-built-tests),yes) -LOCALES := cs_CZ.UTF-8 de_DE.UTF-8 en_US.ISO-8859-1 tr_TR.UTF-8 +LOCALES := cs_CZ.UTF-8 de_DE.UTF-8 en_US.ISO-8859-1 tr_TR.UTF-8 \ + tr_TR.ISO-8859-9 include ../gen-locales.mk $(objpfx)bug-strtod2.out: $(gen-locales) @@ -131,6 +133,7 @@ $(objpfx)tst-strtod.out: $(gen-locales) $(objpfx)tst-strtod3.out: $(gen-locales) $(objpfx)tst-strtod4.out: $(gen-locales) $(objpfx)tst-strtod5.out: $(gen-locales) +$(objpfx)tst-strtol-locale.out: $(gen-locales) endif # Testdir has to be named stdlib and needs to be writable |