aboutsummaryrefslogtreecommitdiff
path: root/iconvdata
diff options
context:
space:
mode:
authorUlrich Drepper <drepper@redhat.com>2000-09-18 22:41:47 +0000
committerUlrich Drepper <drepper@redhat.com>2000-09-18 22:41:47 +0000
commit755104edc75c53f4a0e7440334e944ad3c6b32fc (patch)
tree536824a5d458248d7fc12dc94ae882f8fce58871 /iconvdata
parent8a98b84708dd7438c7ee7055b8b1bda983a53fff (diff)
downloadglibc-755104edc75c53f4a0e7440334e944ad3c6b32fc.zip
glibc-755104edc75c53f4a0e7440334e944ad3c6b32fc.tar.gz
glibc-755104edc75c53f4a0e7440334e944ad3c6b32fc.tar.bz2
Update.
2000-09-18 Ulrich Drepper <drepper@redhat.com> * version.h (VERSION): Bump to 2.1.94. * malloc/mtrace.c (mtrace): Mark stream as close on exec. 2000-09-17 Bruno Haible <haible@clisp.cons.org> * iconvdata/utf-16.c (BODY for TO_LOOP): Reject UCS-4 input in the range 0xD800..0xDFFF. * iconvdata/unicode.c (BODY for TO_LOOP): Likewise. (BODY for FROM_LOOP): Likewise. * iconv/gconv_simple.c (ucs2_internal_loop): Likewise. (internal_ucs2_loop): Likewise. (ucs2reverse_internal_loop): Likewise. (internal_ucs2reverse_loop): Likewise. 2000-09-17 Bruno Haible <haible@clisp.cons.org> * iconvdata/utf-16.c (gconv_init): Add missing slashes to encoding names. 2000-09-17 Bruno Haible <haible@clisp.cons.org> * iconvdata/tst-table-from.c (main): Fix test for error on stdout. * iconvdata/tst-table-to.c (main): Likewise. 2000-09-17 Bruno Haible <haible@clisp.cons.org> * iconvdata/iso-ir-165.c (__isoir165_from_tab): Renamed from __isoir165_tab. * iconvdata/cns11643.h (__cns11643l1_to_ucs4_tab): New declaration. * iconvdata/iso-2022-cn-ext.c: Include "cns11643.h". (GB7590_set, GB13132_set, CNS11643_3_set, CNS11643_4_set, CNS11643_5_set, CNS11643_6_set, CNS11643_7_set): Change enum values. (BODY for FROM_LOOP): Fix buffer overrun. Treat CNS11643 plane 3. Return __GCONV_INCOMPLETE_INPUT instead of __GCONV_EMPTY_INPUT. (BODY for TO_LOOP): Fix usage of `set' vs. `used'. Fix typo that caused GB2312 to be used instead of ISO-IR-165. Treat CNS11643 plane 3. Fix shift sequences. Output announcement for SS2 and SS3 encodings when needed. When outputting an announcement, don't clear most other announcements. 2000-09-17 Bruno Haible <haible@clisp.cons.org> * iconvdata/iso-2022-cn.c (BODY for FROM_LOOP): Fix buffer overrun. (BODY for TO_LOOP): Fix usage of `set' vs. `used'. 2000-09-14 Bruno Haible <haible@clisp.cons.org> * intl/Versions: Add bind_textdomain_codeset.
Diffstat (limited to 'iconvdata')
-rw-r--r--iconvdata/cns11643.h3
-rw-r--r--iconvdata/iso-2022-cn-ext.c251
-rw-r--r--iconvdata/iso-2022-cn.c25
-rw-r--r--iconvdata/iso-ir-165.c2
-rw-r--r--iconvdata/tst-table-from.c2
-rw-r--r--iconvdata/tst-table-to.c2
-rw-r--r--iconvdata/unicode.c32
-rw-r--r--iconvdata/utf-16.c28
8 files changed, 240 insertions, 105 deletions
diff --git a/iconvdata/cns11643.h b/iconvdata/cns11643.h
index b57aa9d..8c73c06 100644
--- a/iconvdata/cns11643.h
+++ b/iconvdata/cns11643.h
@@ -20,8 +20,11 @@
#include <stdint.h>
+/* Table for CNS 11643, plane 1 to UCS4 conversion. */
+extern const uint16_t __cns11643l1_to_ucs4_tab[];
/* Table for CNS 11643, plane 2 to UCS4 conversion. */
extern const uint16_t __cns11643l2_to_ucs4_tab[];
+/* Table for CNS 11643, plane 14 to UCS4 conversion. */
extern const uint16_t __cns11643l14_to_ucs4_tab[];
diff --git a/iconvdata/iso-2022-cn-ext.c b/iconvdata/iso-2022-cn-ext.c
index c1bd7ac..32a639a 100644
--- a/iconvdata/iso-2022-cn-ext.c
+++ b/iconvdata/iso-2022-cn-ext.c
@@ -24,6 +24,7 @@
#include <string.h>
#include "gb2312.h"
#include "iso-ir-165.h"
+#include "cns11643.h"
#include "cns11643l1.h"
#include "cns11643l2.h"
@@ -80,41 +81,41 @@ enum
ISO_IR_165_set,
SO_mask = 7,
- GB7589_set = 8,
- GB13131_set = 16,
- CNS11643_2_set = 24,
- SS2_mask = 24,
+ GB7589_set = 1 << 3,
+ GB13131_set = 2 << 3,
+ CNS11643_2_set = 3 << 3,
+ SS2_mask = 3 << 3,
- GB7590_set = 0,
- GB13132_set = 32,
- CNS11643_3_set = 64,
- CNS11643_4_set = 96,
- CNS11643_5_set = 128,
- CNS11643_6_set = 160,
- CNS11643_7_set = 192,
- SS3_mask = 224,
+ GB7590_set = 1 << 5,
+ GB13132_set = 2 << 5,
+ CNS11643_3_set = 3 << 5,
+ CNS11643_4_set = 4 << 5,
+ CNS11643_5_set = 5 << 5,
+ CNS11643_6_set = 6 << 5,
+ CNS11643_7_set = 7 << 5,
+ SS3_mask = 7 << 5,
#define CURRENT_MASK (SO_mask | SS2_mask | SS3_mask)
- GB2312_ann = 256,
- GB12345_ann = 512,
- CNS11643_1_ann = 768,
- ISO_IR_165_ann = 1024,
- SO_ann = 1792,
+ GB2312_ann = 1 << 8,
+ GB12345_ann = 2 << 8,
+ CNS11643_1_ann = 3 << 8,
+ ISO_IR_165_ann = 4 << 8,
+ SO_ann = 7 << 8,
- GB7589_ann = 2048,
- GB13131_ann = 4096,
- CNS11643_2_ann = 6144,
- SS2_ann = 6144,
+ GB7589_ann = 1 << 11,
+ GB13131_ann = 2 << 11,
+ CNS11643_2_ann = 3 << 11,
+ SS2_ann = 3 << 11,
- GB7590_ann = 8192,
- GB13132_ann = 16384,
- CNS11643_3_ann = 24576,
- CNS11643_4_ann = 32768,
- CNS11643_5_ann = 40960,
- CNS11643_6_ann = 49152,
- CNS11643_7_ann = 57344,
- SS3_ann = 57344
+ GB7590_ann = 1 << 13,
+ GB13132_ann = 2 << 13,
+ CNS11643_3_ann = 3 << 13,
+ CNS11643_4_ann = 4 << 13,
+ CNS11643_5_ann = 5 << 13,
+ CNS11643_6_ann = 6 << 13,
+ CNS11643_7_ann = 7 << 13,
+ SS3_ann = 7 << 13
};
@@ -190,16 +191,16 @@ enum
- the initial byte of the SS2 sequence. \
- the initial byte of the SS3 sequence. \
*/ \
- if (inptr + 1 > inend \
+ if (inptr + 2 > inend \
|| (inptr[1] == '$' \
- && (inptr + 2 > inend \
- || (inptr[2] == ')' && inptr + 3 > inend) \
- || (inptr[2] == '*' && inptr + 3 > inend) \
- || (inptr[2] == '+' && inptr + 3 > inend))) \
- || (inptr[1] == SS2_1 && inptr + 3 > inend) \
- || (inptr[1] == SS3_1 && inptr + 3 > inend)) \
+ && (inptr + 3 > inend \
+ || (inptr[2] == ')' && inptr + 4 > inend) \
+ || (inptr[2] == '*' && inptr + 4 > inend) \
+ || (inptr[2] == '+' && inptr + 4 > inend))) \
+ || (inptr[1] == SS2_1 && inptr + 4 > inend) \
+ || (inptr[1] == SS3_1 && inptr + 4 > inend)) \
{ \
- result = __GCONV_EMPTY_INPUT; \
+ result = __GCONV_INCOMPLETE_INPUT; \
break; \
} \
if (inptr[1] == '$' \
@@ -285,17 +286,12 @@ enum
continue; \
} \
\
- if (ch == ESC && (inend - inptr == 1 || inptr[1] == SS2_1)) \
+ if (ch == ESC && inptr[1] == SS2_1) \
{ \
/* This is a character from CNS 11643 plane 2. \
XXX We could test here whether the use of this character \
set was announced. \
XXX Current GB7589 and GB13131 are not supported. */ \
- if (inend - inptr < 4) \
- { \
- result = __GCONV_INCOMPLETE_INPUT; \
- break; \
- } \
inptr += 2; \
ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \
if (ch == __UNKNOWN_10646_CHAR) \
@@ -306,35 +302,53 @@ enum
result = __GCONV_ILLEGAL_INPUT; \
break; \
} \
+ inptr += 2; \
++*irreversible; \
continue; \
} \
} \
- /* Note that we can assume here that at least bytes are available if \
+ /* Note that we can assume here that at least 4 bytes are available if \
the first byte is ESC since otherwise the first if would have been \
true. */ \
else if (ch == ESC && inptr[1] == SS3_1) \
{ \
/* This is a character from CNS 11643 plane 3 or higher. \
- XXX Current GB7590 and GB13132 are not supported. */ \
- if (inend - inptr < 4) \
+ XXX Currently GB7590 and GB13132 are not supported. */ \
+ char buf[3]; \
+ const char *tmp = buf; \
+ \
+ buf[1] = inptr[2]; \
+ buf[2] = inptr[3]; \
+ switch (ann & SS3_ann) \
{ \
- result = __GCONV_INCOMPLETE_INPUT; \
+ case CNS11643_3_ann: \
+ /* CNS 11643 plane 3 is part of the old CNS 11643 plane 14. */ \
+ if (buf[1] < 0x62 || (buf[1] == 0x62 && buf[2] <= 0x45)) \
+ { \
+ buf[0] = 0x2e; \
+ ch = cns11643_to_ucs4 (&tmp, 3, 0); \
+ } \
+ else \
+ ch = __UNKNOWN_10646_CHAR; \
+ break; \
+ default: \
+ /* XXX Currently planes 4 to 7 are not supported. */ \
+ ch = __UNKNOWN_10646_CHAR; \
break; \
} \
- inptr += 2; \
- ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \
if (ch == __UNKNOWN_10646_CHAR) \
{ \
if (! ignore_errors_p ()) \
{ \
- inptr -= 2; \
result = __GCONV_ILLEGAL_INPUT; \
break; \
} \
+ inptr += 4; \
++*irreversible; \
continue; \
} \
+ assert (tmp == buf + 3); \
+ inptr += 4; \
} \
else if (set == ASCII_set) \
{ \
@@ -361,7 +375,7 @@ enum
\
if (ch == 0) \
{ \
- result = __GCONV_EMPTY_INPUT; \
+ result = __GCONV_INCOMPLETE_INPUT; \
break; \
} \
else if (ch == __UNKNOWN_10646_CHAR) \
@@ -427,16 +441,16 @@ enum
char buf[2]; \
int used; \
\
- if (set == GB2312_set || ((ann & CNS11643_1_ann) == 0 \
- && (ann & ISO_IR_165_ann) == 0)) \
+ if (set == GB2312_set || ((ann & SO_ann) != CNS11643_1_ann \
+ && (ann & SO_ann) != ISO_IR_165_ann)) \
{ \
written = ucs4_to_gb2312 (ch, buf, 2); \
used = GB2312_set; \
} \
- else if (set == ISO_IR_165_set || (ann & ISO_IR_165_set) != 0) \
+ else if (set == ISO_IR_165_set || (ann & SO_ann) == ISO_IR_165_set) \
{ \
- written = ucs4_to_gb2312 (ch, buf, 2); \
- used = GB2312_set; \
+ written = ucs4_to_isoir165 (ch, buf, 2); \
+ used = ISO_IR_165_set; \
} \
else \
{ \
@@ -454,29 +468,66 @@ enum
used = CNS11643_2_set; \
else \
{ \
- /* Well, see whether we have to change the SO set. */ \
- if (set != GB2312_set) \
- { \
- written = ucs4_to_gb2312 (ch, buf, 2); \
- if (written != __UNKNOWN_10646_CHAR) \
- used = GB2312_set; \
- } \
- if (written == __UNKNOWN_10646_CHAR && set != ISO_IR_165_set) \
- { \
- written = ucs4_to_isoir165 (ch, buf, 2); \
- if (written != __UNKNOWN_10646_CHAR) \
- used = ISO_IR_165_set; \
- } \
- if (written == __UNKNOWN_10646_CHAR && set != CNS11643_1_set) \
- { \
- written = ucs4_to_cns11643l1 (ch, buf, 2); \
- if (written != __UNKNOWN_10646_CHAR) \
- used = CNS11643_1_set; \
- } \
+ char tmpbuf[3]; \
\
- if (written == __UNKNOWN_10646_CHAR) \
+ switch (0) \
{ \
+ default: \
+ /* Well, see whether we have to change the SO set. */ \
+ \
+ if (used != GB2312_set) \
+ { \
+ written = ucs4_to_gb2312 (ch, buf, 2); \
+ if (written != __UNKNOWN_10646_CHAR) \
+ { \
+ used = GB2312_set; \
+ break; \
+ } \
+ } \
+ \
+ if (used != ISO_IR_165_set) \
+ { \
+ written = ucs4_to_isoir165 (ch, buf, 2); \
+ if (written != __UNKNOWN_10646_CHAR) \
+ { \
+ used = ISO_IR_165_set; \
+ break; \
+ } \
+ } \
+ \
+ if (used != CNS11643_1_set) \
+ { \
+ written = ucs4_to_cns11643l1 (ch, buf, 2); \
+ if (written != __UNKNOWN_10646_CHAR) \
+ { \
+ used = CNS11643_1_set; \
+ break; \
+ } \
+ } \
+ \
+ written = ucs4_to_cns11643 (ch, tmpbuf, 3); \
+ if (written == 3 && tmpbuf[0] != 1 && tmpbuf[0] != 2) \
+ { \
+ buf[0] = tmpbuf[1]; \
+ buf[1] = tmpbuf[2]; \
+ written = 2; \
+ /* CNS 11643 plane 3 is part of the old CNS 11643 \
+ plane 14. \
+ XXX Currently planes 4 to 7 are not supported. */ \
+ if (tmpbuf[0] == 14 \
+ && (tmpbuf[1] < 0x62 \
+ || (tmpbuf[1] == 0x62 && tmpbuf[2] <= 0x45))) \
+ { \
+ used = CNS11643_3_set; \
+ break; \
+ } \
+ } \
+ \
/* Even this does not work. Error. */ \
+ used = ASCII_set; \
+ } \
+ if (used == ASCII_set) \
+ { \
STANDARD_ERR_HANDLER (4); \
} \
} \
@@ -488,7 +539,7 @@ enum
{ \
/* First see whether we announced that we use this \
character set. */ \
- if ((ann & (2 << used)) == 0) \
+ if ((used & SO_mask) != 0 && (ann & SO_ann) != (used << 8)) \
{ \
const char *escseq; \
\
@@ -499,18 +550,39 @@ enum
} \
\
assert (used >= 1 && used <= 4); \
- escseq = "\e$)A\e$)G\e$*H\e$)E" + (used - 1) * 4; \
+ escseq = ")A\0\0)G)E" + (used - 1) * 2; \
+ *outptr++ = ESC; \
+ *outptr++ = '$'; \
+ *outptr++ = *escseq++; \
+ *outptr++ = *escseq++; \
+ \
+ ann = (ann & ~SO_ann) | (used << 8); \
+ } \
+ else if ((used & SS2_mask) != 0 && (ann & SS2_ann) != (used << 8))\
+ { \
+ const char *escseq; \
+ \
+ assert (used == CNS11643_2_set); /* XXX */ \
+ escseq = "*H"; \
+ *outptr++ = ESC; \
+ *outptr++ = '$'; \
*outptr++ = *escseq++; \
*outptr++ = *escseq++; \
+ \
+ ann = (ann & ~SS2_ann) | (used << 8); \
+ } \
+ else if ((used & SS3_mask) != 0 && (ann & SS3_ann) != (used << 8))\
+ { \
+ const char *escseq; \
+ \
+ assert ((used >> 5) >= 3 && (used >> 5) <= 7); \
+ escseq = "+I+J+K+L+M" + ((used >> 5) - 3) * 2; \
+ *outptr++ = ESC; \
+ *outptr++ = '$'; \
*outptr++ = *escseq++; \
*outptr++ = *escseq++; \
\
- if (used == GB2312_set) \
- ann = (ann & CNS11643_2_ann) | GB2312_ann; \
- else if (used == CNS11643_1_set) \
- ann = (ann & CNS11643_2_ann) | CNS11643_1_ann; \
- else \
- ann |= CNS11643_2_ann; \
+ ann = (ann & ~SS3_ann) | (used << 8); \
} \
\
if (used == CNS11643_2_set) \
@@ -523,6 +595,16 @@ enum
*outptr++ = SS2_0; \
*outptr++ = SS2_1; \
} \
+ else if (used >= CNS11643_3_set && used <= CNS11643_7_set) \
+ { \
+ if (outptr + 2 > outend) \
+ { \
+ result = __GCONV_FULL_OUTPUT; \
+ break; \
+ } \
+ *outptr++ = SS3_0; \
+ *outptr++ = SS3_1; \
+ } \
else \
{ \
/* We only have to emit something if currently ASCII is \
@@ -555,6 +637,7 @@ enum
\
*outptr++ = buf[0]; \
*outptr++ = buf[1]; \
+ set = used; \
} \
\
/* Now that we wrote the output increment the input pointer. */ \
diff --git a/iconvdata/iso-2022-cn.c b/iconvdata/iso-2022-cn.c
index d45ed6b..6040e12 100644
--- a/iconvdata/iso-2022-cn.c
+++ b/iconvdata/iso-2022-cn.c
@@ -141,15 +141,15 @@ enum
line; we can simply ignore them \
- the initial byte of the SS2 sequence. \
*/ \
- if (__builtin_expect (inptr + 1 > inend, 0) \
+ if (__builtin_expect (inptr + 2 > inend, 0) \
|| (inptr[1] == '$' \
- && (__builtin_expect (inptr + 2 > inend, 0) \
+ && (__builtin_expect (inptr + 3 > inend, 0) \
|| (inptr[2] == ')' \
- && __builtin_expect (inptr + 3 > inend, 0)) \
+ && __builtin_expect (inptr + 4 > inend, 0)) \
|| (inptr[2] == '*' \
- && __builtin_expect (inptr + 3 > inend, 0)))) \
+ && __builtin_expect (inptr + 4 > inend, 0)))) \
|| (inptr[1] == SS2_1 \
- && __builtin_expect (inptr + 3 > inend, 0))) \
+ && __builtin_expect (inptr + 4 > inend, 0))) \
{ \
result = __GCONV_INCOMPLETE_INPUT; \
break; \
@@ -313,14 +313,14 @@ enum
else \
{ \
/* Well, see whether we have to change the SO set. */ \
- if (set == GB2312_set) \
+ if (used == GB2312_set) \
written = ucs4_to_cns11643l1 (ch, buf, 2); \
else \
written = ucs4_to_gb2312 (ch, buf, 2); \
\
if (__builtin_expect (written, 0) != __UNKNOWN_10646_CHAR) \
/* Oh well, then switch SO. */ \
- used = GB2312_set + CNS11643_1_set - set; \
+ used = GB2312_set + CNS11643_1_set - used; \
else \
{ \
/* Even this does not work. Error. */ \
@@ -335,7 +335,7 @@ enum
{ \
/* First see whether we announced that we use this \
character set. */ \
- if ((ann & (2 << used)) == 0) \
+ if ((ann & (16 << (used >> 3))) == 0) \
{ \
const char *escseq; \
\
@@ -345,10 +345,10 @@ enum
break; \
} \
\
- assert (used >= 1 && used <= 3); \
- escseq = "\e$)A\e$)G\e$*H" + (used - 1) * 4; \
- *outptr++ = *escseq++; \
- *outptr++ = *escseq++; \
+ assert ((used >> 3) >= 1 && (used >> 3) <= 3); \
+ escseq = ")A)G*H" + ((used >> 3) - 1) * 2; \
+ *outptr++ = ESC; \
+ *outptr++ = '$'; \
*outptr++ = *escseq++; \
*outptr++ = *escseq++; \
\
@@ -402,6 +402,7 @@ enum
\
*outptr++ = buf[0]; \
*outptr++ = buf[1]; \
+ set = used; \
} \
\
/* Now that we wrote the output increment the input pointer. */ \
diff --git a/iconvdata/iso-ir-165.c b/iconvdata/iso-ir-165.c
index cbb4797..529f7ab 100644
--- a/iconvdata/iso-ir-165.c
+++ b/iconvdata/iso-ir-165.c
@@ -546,7 +546,7 @@ const struct gap __isoir165_from_idx[] =
};
-const char __isoir165_tab[29852] =
+const char __isoir165_from_tab[29852] =
"\x2a\x21" "\x2a\x22" "\x2a\x23" "\x21\x67" "\x2a\x25" "\x2a\x26" "\x2a\x27"
"\x2a\x28" "\x2a\x29" "\x2a\x2a" "\x2a\x2b" "\x2a\x2c" "\x2a\x2d" "\x2a\x2e"
"\x2a\x2f" "\x2a\x30" "\x2a\x31" "\x2a\x32" "\x2a\x33" "\x2a\x34" "\x2a\x35"
diff --git a/iconvdata/tst-table-from.c b/iconvdata/tst-table-from.c
index 92a562d..fb4934f 100644
--- a/iconvdata/tst-table-from.c
+++ b/iconvdata/tst-table-from.c
@@ -216,7 +216,7 @@ main (int argc, char *argv[])
exit (1);
}
- if (ferror (stdin) || ferror (stdout))
+ if (ferror (stdin) || fflush (stdout) || ferror (stdout))
{
fprintf (stderr, "I/O error\n");
exit (1);
diff --git a/iconvdata/tst-table-to.c b/iconvdata/tst-table-to.c
index 329ba4a..f154116 100644
--- a/iconvdata/tst-table-to.c
+++ b/iconvdata/tst-table-to.c
@@ -97,7 +97,7 @@ main (int argc, char *argv[])
exit (1);
}
- if (ferror (stdin) || ferror (stdout))
+ if (ferror (stdin) || fflush (stdout) || ferror (stdout))
{
fprintf (stderr, "I/O error\n");
exit (1);
diff --git a/iconvdata/unicode.c b/iconvdata/unicode.c
index 52c2c9d..b8ea905 100644
--- a/iconvdata/unicode.c
+++ b/iconvdata/unicode.c
@@ -154,6 +154,23 @@ gconv_end (struct __gconv_step *data)
{ \
STANDARD_ERR_HANDLER (4); \
} \
+ else if (__builtin_expect (c >= 0xd800 && c < 0xe000, 0)) \
+ { \
+ /* Surrogate characters in UCS-4 input are not valid. \
+ We must catch this, because the UCS-2 output might be \
+ interpreted as UTF-16 by other programs. If we let \
+ surrogates pass through, attackers could make a security \
+ hole exploit by synthesizing any desired plane 1-16 \
+ character. */ \
+ if (! ignore_errors_p ()) \
+ { \
+ result = __GCONV_ILLEGAL_INPUT; \
+ break; \
+ } \
+ inptr += 4; \
+ ++*irreversible; \
+ continue; \
+ } \
else \
{ \
put16 (outptr, c); \
@@ -179,11 +196,26 @@ gconv_end (struct __gconv_step *data)
if (swap) \
u1 = bswap_16 (u1); \
\
+ if (__builtin_expect (u1 >= 0xd800 && u1 < 0xe000, 0)) \
+ { \
+ /* Surrogate characters in UCS-2 input are not valid. Reject \
+ them. (Catching this here is not security relevant.) */ \
+ if (! ignore_errors_p ()) \
+ { \
+ result = __GCONV_ILLEGAL_INPUT; \
+ break; \
+ } \
+ inptr += 2; \
+ ++*irreversible; \
+ continue; \
+ } \
+ \
put32 (outptr, u1); \
\
inptr += 2; \
outptr += 4; \
}
+#define LOOP_NEED_FLAGS
#define EXTRA_LOOP_DECLS \
, int swap
#include <iconv/loop.c>
diff --git a/iconvdata/utf-16.c b/iconvdata/utf-16.c
index 4b7fefa..aa0d00c 100644
--- a/iconvdata/utf-16.c
+++ b/iconvdata/utf-16.c
@@ -109,32 +109,32 @@ gconv_init (struct __gconv_step *step)
enum variant var = illegal_var;
int result;
- if (__strcasecmp (step->__from_name, "UTF-16") == 0)
+ if (__strcasecmp (step->__from_name, "UTF-16//") == 0)
{
dir = from_utf16;
var = UTF_16;
}
- else if (__strcasecmp (step->__to_name, "UTF-16") == 0)
+ else if (__strcasecmp (step->__to_name, "UTF-16//") == 0)
{
dir = to_utf16;
var = UTF_16;
}
- else if (__strcasecmp (step->__from_name, "UTF-16BE") == 0)
+ else if (__strcasecmp (step->__from_name, "UTF-16BE//") == 0)
{
dir = from_utf16;
var = UTF_16BE;
}
- else if (__strcasecmp (step->__to_name, "UTF-16BE") == 0)
+ else if (__strcasecmp (step->__to_name, "UTF-16BE//") == 0)
{
dir = to_utf16;
var = UTF_16BE;
}
- else if (__strcasecmp (step->__from_name, "UTF-16LE") == 0)
+ else if (__strcasecmp (step->__from_name, "UTF-16LE//") == 0)
{
dir = from_utf16;
var = UTF_16LE;
}
- else if (__strcasecmp (step->__to_name, "UTF-16LE") == 0)
+ else if (__strcasecmp (step->__to_name, "UTF-16LE//") == 0)
{
dir = to_utf16;
var = UTF_16LE;
@@ -196,6 +196,22 @@ gconv_end (struct __gconv_step *data)
{ \
uint32_t c = get32 (inptr); \
\
+ if (__builtin_expect (c >= 0xd800 && c < 0xe000, 0)) \
+ { \
+ /* Surrogate characters in UCS-4 input are not valid. \
+ We must catch this. If we let surrogates pass through, \
+ attackers could make a security hole exploit by \
+ synthesizing any desired plane 1-16 character. */ \
+ if (! ignore_errors_p ()) \
+ { \
+ result = __GCONV_ILLEGAL_INPUT; \
+ break; \
+ } \
+ inptr += 4; \
+ ++*irreversible; \
+ continue; \
+ } \
+ \
if (swap) \
{ \
if (__builtin_expect (c, 0) >= 0x10000) \