From 29132b91519572e6076003735cd6f718ff2e1369 Mon Sep 17 00:00:00 2001 From: Ulrich Drepper Date: Wed, 5 Dec 2001 19:19:20 +0000 Subject: Update. * iconvdata/ibm1129.h: Remove duplicate mappings. * iconvdata/ibm937.c: Handle overflow errors. Handle new tables. * iconvdata/ibm937.h: Reorganize table to safe a lot of space. Patch by Masahide Washizawa . * timezone/zic.c: Fix handling of turnaround times. Patch by Arthur David Olson . 2001-12-02 Moshe Olshansky * sysdeps/ieee754/dbl-64/e_remainder.c (__ieee754_remainder): Fix overflow problem. 2001-12-05 Ulrich Drepper * posix/regex.c: For use outside glibc defined bounded pointer macros here. Patch by Jim Meyering . --- iconvdata/ibm937.c | 94 ++++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 70 insertions(+), 24 deletions(-) (limited to 'iconvdata/ibm937.c') diff --git a/iconvdata/ibm937.c b/iconvdata/ibm937.c index 39468aa0dd..be1d4b92d5 100644 --- a/iconvdata/ibm937.c +++ b/iconvdata/ibm937.c @@ -1,4 +1,4 @@ -/* Conversion to and from IBM937. +/* Conversion from and to IBM937. Copyright (C) 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Masahide Washizawa , 2000. @@ -95,6 +95,7 @@ enum #define BODY \ { \ uint32_t ch = *inptr; \ + uint32_t res; \ \ if (__builtin_expect (ch, 0) == SO) \ { \ @@ -123,9 +124,9 @@ enum \ if (curcs == sb) \ { \ - /* Use the UCS4 table for single byte. */ \ - ch = __ibm937sb_to_ucs4[ch]; \ - if (__builtin_expect (ch, L'\1') == L'\0' && *inptr != '\0') \ + /* Use the IBM937 table for single byte. */ \ + res = __ibm937sb_to_ucs4[ch]; \ + if (__builtin_expect (res, L'\1') == L'\0' && ch != '\0') \ { \ /* This is an illegal character. */ \ if (! ignore_errors_p ()) \ @@ -137,19 +138,34 @@ enum } \ else \ { \ - put32 (outptr, ch); \ + put32 (outptr, res); \ outptr += 4; \ } \ ++inptr; \ } \ else \ { \ - /* Use the IBM937 table for double byte. */ \ + const struct gap *rp2 = __ibm937db_to_ucs4_idx; \ \ assert (curcs == db); \ \ - ch = ibm937db_to_ucs4(inptr[0], inptr[1]); \ - if (__builtin_expect (ch, L'\1') == L'\0' && *inptr != '\0') \ + /* Use the IBM937 table for double byte. */ \ + if (__builtin_expect (inptr + 1 >= inend, 0)) \ + { \ + /* The second character is not available. \ + Store the intermediate result. */ \ + result = __GCONV_INCOMPLETE_INPUT; \ + break; \ + } \ + \ + ch = (ch * 0x100) + inptr[1]; \ + while (ch > rp2->end) \ + ++rp2; \ + \ + if (__builtin_expect (rp2 == NULL, 0) \ + || __builtin_expect (ch < rp2->start, 0) \ + || (res = __ibm937db_to_ucs4[ch + rp2->idx], \ + __builtin_expect (res, L'\1') == L'\0' && ch != '\0')) \ { \ /* This is an illegal character. */ \ if (! ignore_errors_p ()) \ @@ -161,7 +177,7 @@ enum } \ else \ { \ - put32 (outptr, ch); \ + put32 (outptr, res); \ outptr += 4; \ } \ inptr += 2; \ @@ -181,22 +197,40 @@ enum #define BODY \ { \ uint32_t ch = get32 (inptr); \ + const struct gap *rp1 = __ucs4_to_ibm937sb_idx; \ + const struct gap *rp2 = __ucs4_to_ibm937db_idx; \ const char *cp; \ \ - /* Use the UCS4 table for single byte. */ \ - if (__builtin_expect (ch >= (sizeof (__ucs4_to_ibm937sb) \ - / sizeof (__ucs4_to_ibm937sb[0])), 0) \ - || (cp = __ucs4_to_ibm937sb[ch], \ - __builtin_expect (cp[0], '\1') == '\0' && ch != 0)) \ + if (__builtin_expect (ch >= 0xffff, 0)) \ { \ - /* Use the UCS4 table for double byte. */ \ - cp = __ucs4_to_ibm937db[ch]; \ - if (__builtin_expect (ch >= (sizeof (__ucs4_to_ibm937db) \ - / sizeof (__ucs4_to_ibm937db[0])), 0) \ - || __builtin_expect (cp[0], '\1') == '\0') \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ + if (! ignore_errors_p ()) \ { \ - UNICODE_TAG_HANDLER (ch, 4); \ + result = __GCONV_ILLEGAL_INPUT; \ + break; \ + } \ + ++*irreversible; \ + inptr += 4; \ + continue; \ + } \ \ + while (ch > rp1->end) \ + ++rp1; \ + \ + /* Use the UCS4 table for single byte. */ \ + if (__builtin_expect (ch < rp1->start, 0) \ + || (cp = __ucs4_to_ibm937sb[ch + rp1->idx], \ + __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0')) \ + { \ + /* Use the UCS4 table for double byte. */ \ + while (ch > rp2->end) \ + ++rp2; \ + \ + if (__builtin_expect (ch < rp2->start, 0) \ + || (cp = __ucs4_to_ibm937db[ch + rp2->idx], \ + __builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0')) \ + { \ /* This is an illegal character. */ \ if (! ignore_errors_p ()) \ { \ @@ -209,10 +243,16 @@ enum { \ if (curcs == sb) \ { \ + if (__builtin_expect (outptr + 1 > outend, 0)) \ + { \ + result = __GCONV_FULL_OUTPUT; \ + break; \ + } \ *outptr++ = SO; \ curcs = db; \ } \ - if (__builtin_expect (outptr + 1 >= outend, 0)) \ + \ + if (__builtin_expect (outptr + 2 > outend, 0)) \ { \ result = __GCONV_FULL_OUTPUT; \ break; \ @@ -225,15 +265,21 @@ enum { \ if (curcs == db) \ { \ - *outptr++ = SI; \ - curcs = sb; \ - if (__builtin_expect (outptr == outend, 0)) \ + if (__builtin_expect (outptr + 1 > outend, 0)) \ { \ result = __GCONV_FULL_OUTPUT; \ break; \ } \ + *outptr++ = SI; \ + } \ + \ + if (__builtin_expect (outptr + 1 > outend, 0)) \ + { \ + result = __GCONV_FULL_OUTPUT; \ + break; \ } \ *outptr++ = cp[0]; \ + curcs = sb; \ } \ \ /* Now that we wrote the output increment the input pointer. */ \ -- cgit v1.2.3