|
 |
147e83 |
From cddea07761373ce92dc75e8306212d71fa2043ba Mon Sep 17 00:00:00 2001
|
|
|
147e83 |
From: Stefan Liebler <stli@linux.vnet.ibm.com>
|
|
|
147e83 |
Date: Mon, 7 Nov 2016 17:18:39 +0100
|
|
|
147e83 |
Subject: [PATCH 15/17] S390: Fix utf32 to utf16 handling of low surrogates
|
|
|
147e83 |
(disable cu42).
|
|
|
147e83 |
|
|
|
147e83 |
Upstream commit a42a95c43133d69b1108f582cffa0f6986a9c3da
|
|
|
147e83 |
|
|
|
147e83 |
According to the latest Unicode standard, a conversion from/to UTF-xx has
|
|
|
147e83 |
to report an error if the character value is in range of an utf16 surrogate
|
|
|
147e83 |
(0xd800..0xdfff). See https://sourceware.org/ml/libc-help/2015-12/msg00015.html.
|
|
|
147e83 |
|
|
|
147e83 |
Thus the cu42 instruction, which converts from utf32 to utf16, has to be
|
|
|
147e83 |
disabled because it does not report an error in case of a value in range of
|
|
|
147e83 |
a low surrogate (0xdc00..0xdfff). The etf3eh variant is removed and the c,
|
|
|
147e83 |
vector variant is adjusted to handle the value in range of an utf16 low
|
|
|
147e83 |
surrogate correctly.
|
|
|
147e83 |
|
|
|
147e83 |
ChangeLog:
|
|
|
147e83 |
|
|
|
147e83 |
* sysdeps/s390/utf16-utf32-z9.c: Disable cu42 instruction and report
|
|
|
147e83 |
an error in case of a value in range of an utf16 low surrogate.
|
|
|
147e83 |
---
|
|
|
147e83 |
sysdeps/s390/utf16-utf32-z9.c | 155 +++++++++++++++++-------------------------
|
|
|
147e83 |
1 file changed, 62 insertions(+), 93 deletions(-)
|
|
|
147e83 |
|
|
|
147e83 |
diff --git a/sysdeps/s390/utf16-utf32-z9.c b/sysdeps/s390/utf16-utf32-z9.c
|
|
|
147e83 |
index 8d42ab8..5d2ac44 100644
|
|
|
147e83 |
--- a/sysdeps/s390/utf16-utf32-z9.c
|
|
|
147e83 |
+++ b/sysdeps/s390/utf16-utf32-z9.c
|
|
|
147e83 |
@@ -145,42 +145,6 @@ gconv_end (struct __gconv_step *data)
|
|
|
147e83 |
free (data->__data);
|
|
|
147e83 |
}
|
|
|
147e83 |
|
|
|
147e83 |
-/* The macro for the hardware loop. This is used for both
|
|
|
147e83 |
- directions. */
|
|
|
147e83 |
-#define HARDWARE_CONVERT(INSTRUCTION) \
|
|
|
147e83 |
- { \
|
|
|
147e83 |
- register const unsigned char* pInput __asm__ ("8") = inptr; \
|
|
|
147e83 |
- register size_t inlen __asm__ ("9") = inend - inptr; \
|
|
|
147e83 |
- register unsigned char* pOutput __asm__ ("10") = outptr; \
|
|
|
147e83 |
- register size_t outlen __asm__("11") = outend - outptr; \
|
|
|
147e83 |
- unsigned long cc = 0; \
|
|
|
147e83 |
- \
|
|
|
147e83 |
- __asm__ __volatile__ (".machine push \n\t" \
|
|
|
147e83 |
- ".machine \"z9-109\" \n\t" \
|
|
|
147e83 |
- ".machinemode \"zarch_nohighgprs\"\n\t" \
|
|
|
147e83 |
- "0: " INSTRUCTION " \n\t" \
|
|
|
147e83 |
- ".machine pop \n\t" \
|
|
|
147e83 |
- " jo 0b \n\t" \
|
|
|
147e83 |
- " ipm %2 \n" \
|
|
|
147e83 |
- : "+a" (pOutput), "+a" (pInput), "+d" (cc), \
|
|
|
147e83 |
- "+d" (outlen), "+d" (inlen) \
|
|
|
147e83 |
- : \
|
|
|
147e83 |
- : "cc", "memory"); \
|
|
|
147e83 |
- \
|
|
|
147e83 |
- inptr = pInput; \
|
|
|
147e83 |
- outptr = pOutput; \
|
|
|
147e83 |
- cc >>= 28; \
|
|
|
147e83 |
- \
|
|
|
147e83 |
- if (cc == 1) \
|
|
|
147e83 |
- { \
|
|
|
147e83 |
- result = __GCONV_FULL_OUTPUT; \
|
|
|
147e83 |
- } \
|
|
|
147e83 |
- else if (cc == 2) \
|
|
|
147e83 |
- { \
|
|
|
147e83 |
- result = __GCONV_ILLEGAL_INPUT; \
|
|
|
147e83 |
- } \
|
|
|
147e83 |
- }
|
|
|
147e83 |
-
|
|
|
147e83 |
#define PREPARE_LOOP \
|
|
|
147e83 |
enum direction dir = ((struct utf16_data *) step->__data)->dir; \
|
|
|
147e83 |
int emit_bom = ((struct utf16_data *) step->__data)->emit_bom; \
|
|
|
147e83 |
@@ -310,7 +274,7 @@ gconv_end (struct __gconv_step *data)
|
|
|
147e83 |
" slgr %[R_OUTLEN],%[R_TMP3]\n\t" \
|
|
|
147e83 |
/* Calculate remaining uint16_t values in loaded vrs. */ \
|
|
|
147e83 |
"12: lghi %[R_TMP2],16\n\t" \
|
|
|
147e83 |
- " sgr %[R_TMP2],%[R_TMP]\n\t" \
|
|
|
147e83 |
+ " slgr %[R_TMP2],%[R_TMP]\n\t" \
|
|
|
147e83 |
" srl %[R_TMP2],1\n\t" \
|
|
|
147e83 |
" llh %[R_TMP],0(%[R_IN])\n\t" \
|
|
|
147e83 |
" aghi %[R_OUTLEN],-4\n\t" \
|
|
|
147e83 |
@@ -437,7 +401,7 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
uint32_t c = get32 (inptr); \
|
|
|
147e83 |
\
|
|
|
147e83 |
if (__builtin_expect (c <= 0xd7ff, 1) \
|
|
|
147e83 |
- || (c >=0xdc00 && c <= 0xffff)) \
|
|
|
147e83 |
+ || (c > 0xdfff && c <= 0xffff)) \
|
|
|
147e83 |
{ \
|
|
|
147e83 |
/* Two UTF-16 chars. */ \
|
|
|
147e83 |
put16 (outptr, c); \
|
|
|
147e83 |
@@ -475,29 +439,10 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
inptr += 4; \
|
|
|
147e83 |
}
|
|
|
147e83 |
|
|
|
147e83 |
-#define BODY_TO_ETF3EH \
|
|
|
147e83 |
- { \
|
|
|
147e83 |
- HARDWARE_CONVERT ("cu42 %0, %1"); \
|
|
|
147e83 |
- \
|
|
|
147e83 |
- if (__glibc_likely (inptr == inend) \
|
|
|
147e83 |
- || result == __GCONV_FULL_OUTPUT) \
|
|
|
147e83 |
- break; \
|
|
|
147e83 |
- \
|
|
|
147e83 |
- if (inptr + 4 > inend) \
|
|
|
147e83 |
- { \
|
|
|
147e83 |
- result = __GCONV_INCOMPLETE_INPUT; \
|
|
|
147e83 |
- break; \
|
|
|
147e83 |
- } \
|
|
|
147e83 |
- \
|
|
|
147e83 |
- STANDARD_TO_LOOP_ERR_HANDLER (4); \
|
|
|
147e83 |
- }
|
|
|
147e83 |
-
|
|
|
147e83 |
#define BODY_TO_VX \
|
|
|
147e83 |
{ \
|
|
|
147e83 |
- register const unsigned char* pInput asm ("8") = inptr; \
|
|
|
147e83 |
- register size_t inlen asm ("9") = inend - inptr; \
|
|
|
147e83 |
- register unsigned char* pOutput asm ("10") = outptr; \
|
|
|
147e83 |
- register size_t outlen asm("11") = outend - outptr; \
|
|
|
147e83 |
+ size_t inlen = inend - inptr; \
|
|
|
147e83 |
+ size_t outlen = outend - outptr; \
|
|
|
147e83 |
unsigned long tmp, tmp2, tmp3; \
|
|
|
147e83 |
asm volatile (".machine push\n\t" \
|
|
|
147e83 |
".machine \"z13\"\n\t" \
|
|
|
147e83 |
@@ -509,8 +454,8 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
CONVERT_32BIT_SIZE_T ([R_OUTLEN]) \
|
|
|
147e83 |
/* Loop which handles UTF-16 chars \
|
|
|
147e83 |
ch < 0xd800 || (ch > 0xdfff && ch < 0x10000). */ \
|
|
|
147e83 |
- "0: clgijl %[R_INLEN],32,20f\n\t" \
|
|
|
147e83 |
- " clgijl %[R_OUTLEN],16,20f\n\t" \
|
|
|
147e83 |
+ "0: clgijl %[R_INLEN],32,2f\n\t" \
|
|
|
147e83 |
+ " clgijl %[R_OUTLEN],16,2f\n\t" \
|
|
|
147e83 |
"1: vlm %%v16,%%v17,0(%[R_IN])\n\t" \
|
|
|
147e83 |
" lghi %[R_TMP2],0\n\t" \
|
|
|
147e83 |
/* Shorten to UTF-16. */ \
|
|
|
147e83 |
@@ -526,9 +471,15 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
" aghi %[R_INLEN],-32\n\t" \
|
|
|
147e83 |
" aghi %[R_OUTLEN],-16\n\t" \
|
|
|
147e83 |
" la %[R_OUT],16(%[R_OUT])\n\t" \
|
|
|
147e83 |
- " clgijl %[R_INLEN],32,20f\n\t" \
|
|
|
147e83 |
- " clgijl %[R_OUTLEN],16,20f\n\t" \
|
|
|
147e83 |
+ " clgijl %[R_INLEN],32,2f\n\t" \
|
|
|
147e83 |
+ " clgijl %[R_OUTLEN],16,2f\n\t" \
|
|
|
147e83 |
" j 1b\n\t" \
|
|
|
147e83 |
+ /* Calculate remaining uint32_t values in inptr. */ \
|
|
|
147e83 |
+ "2: \n\t" \
|
|
|
147e83 |
+ " clgije %[R_INLEN],0,99f\n\t" \
|
|
|
147e83 |
+ " clgijl %[R_INLEN],4,92f\n\t" \
|
|
|
147e83 |
+ " srlg %[R_TMP2],%[R_INLEN],2\n\t" \
|
|
|
147e83 |
+ " j 20f\n\t" \
|
|
|
147e83 |
/* Setup to check for ch >= 0xd800 && ch <= 0xdfff \
|
|
|
147e83 |
and check for ch >= 0x10000. (v30, v31) */ \
|
|
|
147e83 |
"9: .long 0xd800,0xdfff,0x10000,0x10000\n\t" \
|
|
|
147e83 |
@@ -540,21 +491,59 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
" agr %[R_TMP],%[R_TMP2]\n\t" \
|
|
|
147e83 |
" srlg %[R_TMP3],%[R_TMP],1\n\t" /* Number of out bytes. */ \
|
|
|
147e83 |
" ahik %[R_TMP2],%[R_TMP3],-1\n\t" /* Highest index to store. */ \
|
|
|
147e83 |
- " jl 20f\n\t" \
|
|
|
147e83 |
+ " jl 12f\n\t" \
|
|
|
147e83 |
" vstl %%v18,%[R_TMP2],0(%[R_OUT])\n\t" \
|
|
|
147e83 |
/* Update pointers. */ \
|
|
|
147e83 |
" la %[R_IN],0(%[R_TMP],%[R_IN])\n\t" \
|
|
|
147e83 |
" slgr %[R_INLEN],%[R_TMP]\n\t" \
|
|
|
147e83 |
" la %[R_OUT],0(%[R_TMP3],%[R_OUT])\n\t" \
|
|
|
147e83 |
" slgr %[R_OUTLEN],%[R_TMP3]\n\t" \
|
|
|
147e83 |
- /* Handles UTF16 surrogates with convert instruction. */ \
|
|
|
147e83 |
- "20: cu42 %[R_OUT],%[R_IN]\n\t" \
|
|
|
147e83 |
- " jo 0b\n\t" /* Try vector implemenation again. */ \
|
|
|
147e83 |
- " lochil %[R_RES],%[RES_OUT_FULL]\n\t" /* cc == 1. */ \
|
|
|
147e83 |
- " lochih %[R_RES],%[RES_IN_ILL]\n\t" /* cc == 2. */ \
|
|
|
147e83 |
+ /* Calculate remaining uint32_t values in vrs. */ \
|
|
|
147e83 |
+ "12: lghi %[R_TMP2],8\n\t" \
|
|
|
147e83 |
+ " srlg %[R_TMP3],%[R_TMP3],1\n\t" \
|
|
|
147e83 |
+ " slgr %[R_TMP2],%[R_TMP3]\n\t" \
|
|
|
147e83 |
+ /* Handle remaining UTF-32 characters. */ \
|
|
|
147e83 |
+ "20: l %[R_TMP],0(%[R_IN])\n\t" \
|
|
|
147e83 |
+ " aghi %[R_INLEN],-4\n\t" \
|
|
|
147e83 |
+ /* Test if ch is 2byte UTF-16 char. */ \
|
|
|
147e83 |
+ " clfi %[R_TMP],0xffff\n\t" \
|
|
|
147e83 |
+ " jh 21f\n\t" \
|
|
|
147e83 |
+ /* Handle 2 byte UTF16 char. */ \
|
|
|
147e83 |
+ " lgr %[R_TMP3],%[R_TMP]\n\t" \
|
|
|
147e83 |
+ " nilf %[R_TMP],0xf800\n\t" \
|
|
|
147e83 |
+ " clfi %[R_TMP],0xd800\n\t" \
|
|
|
147e83 |
+ " je 91f\n\t" /* Do not accept UTF-16 surrogates. */ \
|
|
|
147e83 |
+ " slgfi %[R_OUTLEN],2\n\t" \
|
|
|
147e83 |
+ " jl 90f \n\t" \
|
|
|
147e83 |
+ " sth %[R_TMP3],0(%[R_OUT])\n\t" \
|
|
|
147e83 |
+ " la %[R_IN],4(%[R_IN])\n\t" \
|
|
|
147e83 |
+ " la %[R_OUT],2(%[R_OUT])\n\t" \
|
|
|
147e83 |
+ " brctg %[R_TMP2],20b\n\t" \
|
|
|
147e83 |
+ " j 0b\n\t" /* Switch to vx-loop. */ \
|
|
|
147e83 |
+ /* Test if ch is 4byte UTF-16 char. */ \
|
|
|
147e83 |
+ "21: clfi %[R_TMP],0x10ffff\n\t" \
|
|
|
147e83 |
+ " jh 91f\n\t" /* ch > 0x10ffff is not allowed! */ \
|
|
|
147e83 |
+ /* Handle 4 byte UTF16 char. */ \
|
|
|
147e83 |
+ " slgfi %[R_OUTLEN],4\n\t" \
|
|
|
147e83 |
+ " jl 90f \n\t" \
|
|
|
147e83 |
+ " slfi %[R_TMP],0x10000\n\t" /* zabcd = uvwxy - 1. */ \
|
|
|
147e83 |
+ " llilf %[R_TMP3],0xd800dc00\n\t" \
|
|
|
147e83 |
+ " la %[R_IN],4(%[R_IN])\n\t" \
|
|
|
147e83 |
+ " risbgn %[R_TMP3],%[R_TMP],38,47,6\n\t" /* High surrogate. */ \
|
|
|
147e83 |
+ " risbgn %[R_TMP3],%[R_TMP],54,63,0\n\t" /* Low surrogate. */ \
|
|
|
147e83 |
+ " st %[R_TMP3],0(%[R_OUT])\n\t" \
|
|
|
147e83 |
+ " la %[R_OUT],4(%[R_OUT])\n\t" \
|
|
|
147e83 |
+ " brctg %[R_TMP2],20b\n\t" \
|
|
|
147e83 |
+ " j 0b\n\t" /* Switch to vx-loop. */ \
|
|
|
147e83 |
+ "92: lghi %[R_RES],%[RES_IN_FULL]\n\t" \
|
|
|
147e83 |
+ " j 99f\n\t" \
|
|
|
147e83 |
+ "91: lghi %[R_RES],%[RES_IN_ILL]\n\t" \
|
|
|
147e83 |
+ " j 99f\n\t" \
|
|
|
147e83 |
+ "90: lghi %[R_RES],%[RES_OUT_FULL]\n\t" \
|
|
|
147e83 |
+ "99: \n\t" \
|
|
|
147e83 |
".machine pop" \
|
|
|
147e83 |
- : /* outputs */ [R_IN] "+a" (pInput) \
|
|
|
147e83 |
- , [R_INLEN] "+d" (inlen), [R_OUT] "+a" (pOutput) \
|
|
|
147e83 |
+ : /* outputs */ [R_IN] "+a" (inptr) \
|
|
|
147e83 |
+ , [R_INLEN] "+d" (inlen), [R_OUT] "+a" (outptr) \
|
|
|
147e83 |
, [R_OUTLEN] "+d" (outlen), [R_TMP] "=a" (tmp) \
|
|
|
147e83 |
, [R_TMP2] "=d" (tmp2), [R_TMP3] "=a" (tmp3) \
|
|
|
147e83 |
, [R_RES] "+d" (result) \
|
|
|
147e83 |
@@ -567,17 +556,10 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
ASM_CLOBBER_VR ("v18") ASM_CLOBBER_VR ("v19") \
|
|
|
147e83 |
ASM_CLOBBER_VR ("v30") ASM_CLOBBER_VR ("v31") \
|
|
|
147e83 |
); \
|
|
|
147e83 |
- inptr = pInput; \
|
|
|
147e83 |
- outptr = pOutput; \
|
|
|
147e83 |
- \
|
|
|
147e83 |
if (__glibc_likely (inptr == inend) \
|
|
|
147e83 |
- || result == __GCONV_FULL_OUTPUT) \
|
|
|
147e83 |
+ || result != __GCONV_ILLEGAL_INPUT) \
|
|
|
147e83 |
break; \
|
|
|
147e83 |
- if (inptr + 4 > inend) \
|
|
|
147e83 |
- { \
|
|
|
147e83 |
- result = __GCONV_INCOMPLETE_INPUT; \
|
|
|
147e83 |
- break; \
|
|
|
147e83 |
- } \
|
|
|
147e83 |
+ \
|
|
|
147e83 |
STANDARD_TO_LOOP_ERR_HANDLER (4); \
|
|
|
147e83 |
}
|
|
|
147e83 |
|
|
|
147e83 |
@@ -590,15 +572,6 @@ strong_alias (__from_utf16_loop_c_single, __from_utf16_loop_single)
|
|
|
147e83 |
#define BODY BODY_TO_C
|
|
|
147e83 |
#include <iconv/loop.c>
|
|
|
147e83 |
|
|
|
147e83 |
-/* Generate loop-function with hardware utf-convert instruction. */
|
|
|
147e83 |
-#define MIN_NEEDED_INPUT MIN_NEEDED_TO
|
|
|
147e83 |
-#define MIN_NEEDED_OUTPUT MIN_NEEDED_FROM
|
|
|
147e83 |
-#define MAX_NEEDED_OUTPUT MAX_NEEDED_FROM
|
|
|
147e83 |
-#define LOOPFCT __to_utf16_loop_etf3eh
|
|
|
147e83 |
-#define LOOP_NEED_FLAGS
|
|
|
147e83 |
-#define BODY BODY_TO_ETF3EH
|
|
|
147e83 |
-#include <iconv/loop.c>
|
|
|
147e83 |
-
|
|
|
147e83 |
#if defined HAVE_S390_VX_ASM_SUPPORT
|
|
|
147e83 |
/* Generate loop-function with hardware vector instructions. */
|
|
|
147e83 |
# define MIN_NEEDED_INPUT MIN_NEEDED_TO
|
|
|
147e83 |
@@ -623,10 +596,6 @@ __to_utf16_loop_resolver (unsigned long int dl_hwcap)
|
|
|
147e83 |
return __to_utf16_loop_vx;
|
|
|
147e83 |
else
|
|
|
147e83 |
#endif
|
|
|
147e83 |
- if (dl_hwcap & HWCAP_S390_ZARCH && dl_hwcap & HWCAP_S390_HIGH_GPRS
|
|
|
147e83 |
- && dl_hwcap & HWCAP_S390_ETF3EH)
|
|
|
147e83 |
- return __to_utf16_loop_etf3eh;
|
|
|
147e83 |
- else
|
|
|
147e83 |
return __to_utf16_loop_c;
|
|
|
147e83 |
}
|
|
|
147e83 |
|
|
|
147e83 |
--
|
|
|
147e83 |
1.8.3.1
|
|
|
147e83 |
|