Add a testcase for the cu41 insn. Update opcode list, NEWS and bugstatus.
#289839 is finally fixed.
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12851 a5019735-40e9-0310-863c-91ae7b9d1cf9
diff --git a/NEWS b/NEWS
index 9dcf525..10bfa1d 100644
--- a/NEWS
+++ b/NEWS
@@ -182,6 +182,7 @@
289656 == 273475 (Add support for AVX instructions)
289699 vgdb connection in relay mode erroneously closed due to buffer overrun
289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
+289839 s390x: Provide support for unicode conversion instructions
289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
290655 Add support for AESKEYGENASSIST instruction
diff --git a/docs/internals/3_7_BUGSTATUS.txt b/docs/internals/3_7_BUGSTATUS.txt
index 80e1a51..a86836f 100644
--- a/docs/internals/3_7_BUGSTATUS.txt
+++ b/docs/internals/3_7_BUGSTATUS.txt
@@ -71,10 +71,6 @@
289836 Unhanded instructions mfatbu and mfatbl
ppc e500 insns
-289839 s390x:Provide support for unicode conversion instruction
- like cu21,cu12,cu42 etc.
- s390 This will not be ready for 3.8.0
-
289912 link_tool_exe_darwin should not use relative paths
unsupported build configuration; low prio
diff --git a/docs/internals/s390-opcodes.csv b/docs/internals/s390-opcodes.csv
index ff6f135..e5bdbd0 100644
--- a/docs/internals/s390-opcodes.csv
+++ b/docs/internals/s390-opcodes.csv
@@ -662,7 +662,7 @@
cu24,"convert utf-16 to utf-32",implemented
cu21,"convert utf-16 to utf-8",implemented
cu42,"convert utf-32 to utf-16",implemented
-cu41,"convert utf-32 to utf-8","not implemented","open bugzilla"
+cu41,"convert utf-32 to utf-8",implemented
cu12,"convert utf-8 to utf-16",implemented
cu14,"convert utf-8 to utf-32",implemented
srstu,"search string unicode","not implemented",
diff --git a/none/tests/s390x/Makefile.am b/none/tests/s390x/Makefile.am
index e32bfd6..399a829 100644
--- a/none/tests/s390x/Makefile.am
+++ b/none/tests/s390x/Makefile.am
@@ -8,7 +8,7 @@
op_exception fgx stck stckf stcke stfle cksm mvcl clcl troo \
trto trot trtt tr tre cij cgij clij clgij crj cgrj clrj clgrj \
cs csg cds cdsg cu21 cu21_1 cu24 cu24_1 cu42 cu12 cu12_1 \
- ex_sig ex_clone cu14 cu14_1
+ ex_sig ex_clone cu14 cu14_1 cu41
check_PROGRAMS = $(INSN_TESTS) \
allexec \
diff --git a/none/tests/s390x/cu41.c b/none/tests/s390x/cu41.c
new file mode 100644
index 0000000..d6a2f7a
--- /dev/null
+++ b/none/tests/s390x/cu41.c
@@ -0,0 +1,259 @@
+#include <stdint.h>
+#include <inttypes.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include <assert.h>
+#include "opcodes.h"
+
+/* The abstracted result of an CU41 insn */
+typedef struct {
+ uint64_t addr1; // target
+ uint64_t len1;
+ uint64_t addr2; // source
+ uint64_t len2;
+ uint32_t cc;
+} cu41_t;
+
+/* Define various input buffers. */
+
+/* 0000 to 00ff: Result is 1 byte for each uint32_t */
+uint32_t pattern1[] = {
+ 0x0000, 0x007f, /* corner cases */
+ 0x0001, 0x007e, 0x0030, 0x005e /* misc */
+};
+
+/* 0080 to 07ff: Result is 2 bytes for each uint32_t */
+uint32_t pattern2[] = {
+ 0x0080, 0x07ff, /* corner cases */
+ 0x0081, 0x07fe, 0x100, 0x333, 0x555, 0x6aa /* misc */
+};
+
+/* 0800 to d7ff: Result is 3 bytes for each uint32_t */
+/* dc00 to ffff: Result is 3 bytes for each uint32_t */
+uint32_t pattern3[] = {
+ 0x0800, 0xd7ff, /* corner cases */
+ 0xdc00, 0xffff, /* corner cases */
+ 0xdc01, 0xfffe, 0xdea0, 0xd00d, 0xe555 /* misc */
+};
+
+/* 10000 to 10ffff: Result is 4 bytes for each uint32_t */
+uint32_t pattern4[] = {
+ 0x10000, 0x10ffff, /* corner cases */
+ 0x10001, 0x10fffe, 0x12345, 0x23456, 0xfedcb /* misc */
+};
+
+/* Invalid UTF-32 character */
+uint32_t invalid[] = {
+ 0x0000d800, 0x0000dbff, /* corner cases */
+ 0x00110000, 0xffffffff, /* corner cases */
+ 0x0000daad, 0x0000d901, 0x0000d8ff, /* misc */
+ 0x00110011, 0x01000000, 0x10000000, 0xdeadbeef /* misc */
+};
+
+/* Mixed bytes */
+uint32_t mixed[] = {
+ 0x00000078 /* 1 byte */,
+ 0x00000111 /* 2 bytes */,
+ 0x00001234 /* 3 bytes */,
+ 0x00040404 /* 4 bytes */,
+};
+
+/* This is the buffer for the converted bytes. */
+uint8_t buff[1000]; /* Large so we con'don't have to worry about it */
+
+void write_and_check(uint32_t *, unsigned, unsigned);
+
+
+static cu41_t
+do_cu41(uint8_t *dst, uint64_t dst_len, uint32_t *src, uint64_t src_len)
+{
+ int cc = 42;
+ cu41_t regs;
+
+ /* build up the register pairs */
+ register uint32_t *source asm("4") = src;
+ register uint64_t source_len asm("5") = src_len;
+ register uint8_t *dest asm("2") = dst;
+ register uint64_t dest_len asm("3") = dst_len;
+
+ asm volatile(
+ CU41(2,4)
+ "ipm %2\n\t"
+ "srl %2,28\n\t"
+ : "+d"(dest), "+d"(source), "=d"(cc),
+ "+d"(source_len), "+d"(dest_len)
+ :
+ : "memory", "cc");
+
+ /* Capture register contents at end of cu41 */
+ regs.addr1 = (uint64_t)dest;
+ regs.len1 = dest_len;
+ regs.addr2 = (uint64_t)source;
+ regs.len2 = source_len;
+ regs.cc = cc;
+
+ return regs;
+}
+
+void
+run_test(uint8_t *dst, uint64_t dst_len, uint32_t *src, uint64_t src_len)
+{
+ int i;
+ cu41_t result;
+
+ result = do_cu41(dst, dst_len, src, src_len);
+
+ // Write out the converted values, if any
+ printf("UTF8: ");
+ if (dst_len - result.len1 == 0)
+ printf(" <none>");
+ else
+ for (i = 0; i < dst_len - result.len1; ++i) {
+ printf(" %02x", dst[i]);
+ }
+ printf("\n");
+
+ printf(" cc = %d\n", result.cc);
+ if (dst != NULL)
+ printf(" dst address difference: %"PRId64, result.addr1 - (uint64_t)dst);
+ printf(" dst len: %"PRId64"\n", result.len1);
+
+ if (src != NULL)
+ printf(" src address difference: %"PRId64, result.addr2 - (uint64_t)src);
+ printf(" src len: %"PRId64"\n", result.len2);
+}
+
+int main()
+{
+ int i;
+
+ /* Length == 0, no memory should be read or written */
+ printf("\n------------- test1 ----------------\n");
+ run_test(NULL, 0, NULL, 0);
+
+ /* Test exhaustion of source length (source bytes are valid) */
+ printf("\n------------- test2.1 ----------------\n");
+
+ /* No character will be written to BUFF, i.e. loop in jitted code
+ is not iterated */
+ run_test(buff, sizeof buff, NULL, 0);
+ run_test(buff, sizeof buff, NULL, 1);
+ run_test(buff, sizeof buff, NULL, 2);
+ run_test(buff, sizeof buff, NULL, 3);
+ run_test(buff, sizeof buff, pattern1, 0);
+ run_test(buff, sizeof buff, pattern1, 1);
+ run_test(buff, sizeof buff, pattern1, 2);
+ run_test(buff, sizeof buff, pattern1, 3);
+
+ printf("\n------------- test2.2 ----------------\n");
+ /* At least one character will be written to BUFF, i.e. loop in jitted
+ code is iterated */
+ run_test(buff, sizeof buff, pattern1, 4); /* 1 utf32 -> 1 1-byte utf8 */
+ run_test(buff, sizeof buff, pattern2, 10); /* 2 utf32 -> 2 2-byte utf8 */
+ run_test(buff, sizeof buff, pattern3, 5); /* 1 utf32 -> 1 3-byte utf8 */
+ run_test(buff, sizeof buff, pattern4, 21); /* 5 utf32 -> 5 4-byte utf8 */
+
+ /* Test exhaustion of destination length (source bytes are valid) */
+ printf("\n------------- test3.1 ----------------\n");
+
+ /* No character will be written to BUFF, i.e. loop in jitted code
+ is not iterated */
+
+ /* Want to write at least 1 byte */
+ run_test(NULL, 0, pattern1, sizeof pattern1);
+
+ /* Want to write at least 2 bytes */
+ run_test(NULL, 0, pattern2, sizeof pattern2);
+ run_test(NULL, 1, pattern2, sizeof pattern2);
+
+ /* Want to write at least 3 bytes */
+ run_test(NULL, 0, pattern3, sizeof pattern3);
+ run_test(NULL, 1, pattern3, sizeof pattern3);
+
+ /* Want to write at least 4 bytes */
+ run_test(NULL, 0, pattern4, sizeof pattern4);
+ run_test(NULL, 1, pattern4, sizeof pattern4);
+ run_test(NULL, 2, pattern4, sizeof pattern4);
+ run_test(NULL, 3, pattern4, sizeof pattern4);
+
+ /* When both operands are exhausted, cc=0 takes precedence.
+ (test1 tests this for len == 0) */
+ printf("\n------------- test4 ----------------\n");
+ run_test(buff, 2, pattern1, 8);
+
+ /* Input contains invalid characters */
+
+ // As conversion stops upon encountering an invalid character, we
+ // need to test each invalid character separately, to make sure it
+ // is recognized as invalid.
+
+ printf("\n------------- test5 ----------------\n");
+ for (i = 0; i < sizeof invalid / 4; ++i) {
+ run_test(buff, sizeof buff, invalid + i, 4);
+ }
+ run_test(buff, 0, invalid, sizeof invalid); // cc = 2
+ run_test(buff, 100, invalid, sizeof invalid);
+
+ /* Convert all pattern buffers */
+ printf("\n------------- test6 ----------------\n");
+ run_test(buff, sizeof buff, pattern1, sizeof pattern1);
+ run_test(buff, sizeof buff, pattern2, sizeof pattern2);
+ run_test(buff, sizeof buff, pattern3, sizeof pattern3);
+ run_test(buff, sizeof buff, pattern4, sizeof pattern4);
+ run_test(buff, sizeof buff, mixed, sizeof mixed);
+
+ /* Make sure we only write the exact number of bytes (and not more) */
+
+ /* Write 1 byte */
+ printf("\n------------- test7.0 ----------------\n");
+ write_and_check(pattern1 + 2, 4, 1);
+
+ /* Write 2 bytes */
+ printf("\n------------- test7.1 ----------------\n");
+ write_and_check(pattern2 + 3, 4, 2);
+
+ /* Write 3 bytes */
+ printf("\n------------- test7.2 ----------------\n");
+ write_and_check(pattern3 + 6, 4, 3);
+
+ /* Write 4 bytes */
+ printf("\n------------- test7.3 ----------------\n");
+ write_and_check(pattern4 + 5, 4, 4);
+
+ return 0;
+}
+
+
+void
+write_and_check_aux(uint32_t *input, unsigned num_input_bytes,
+ unsigned num_expected_output_bytes,
+ unsigned fill_byte)
+{
+ int num_errors, i;
+
+ /* Fill output buffer with FILL_BYTE */
+ memset(buff, fill_byte, sizeof buff);
+
+ /* Execute cu41 */
+ run_test(buff, sizeof buff, input, num_input_bytes);
+
+ /* Make sure the rest of the buffer is unmodified. */
+ num_errors = 0;
+ for (i = num_expected_output_bytes; i < sizeof buff; ++i)
+ if (((unsigned char *)buff)[i] != fill_byte) ++num_errors;
+ if (num_errors)
+ fprintf(stderr, "*** wrote more than %d bytes\n",
+ num_expected_output_bytes);
+}
+
+void
+write_and_check(uint32_t *input, unsigned num_input_bytes,
+ unsigned num_expected_output_bytes)
+{
+ write_and_check_aux(input, num_input_bytes, num_expected_output_bytes, 0x0);
+
+ /* Run again with different fill pattern to make sure we did not write
+ an extra 0x0 byte */
+ write_and_check_aux(input, num_input_bytes, num_expected_output_bytes, 0xFF);
+}
diff --git a/none/tests/s390x/cu41.stderr.exp b/none/tests/s390x/cu41.stderr.exp
new file mode 100644
index 0000000..139597f
--- /dev/null
+++ b/none/tests/s390x/cu41.stderr.exp
@@ -0,0 +1,2 @@
+
+
diff --git a/none/tests/s390x/cu41.stdout.exp b/none/tests/s390x/cu41.stdout.exp
new file mode 100644
index 0000000..c7fa41f
--- /dev/null
+++ b/none/tests/s390x/cu41.stdout.exp
@@ -0,0 +1,218 @@
+
+------------- test1 ----------------
+UTF8: <none>
+ cc = 0
+ dst len: 0
+ src len: 0
+
+------------- test2.1 ----------------
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src len: 0
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src len: 1
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src len: 2
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src len: 3
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 0
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 1
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 2
+UTF8: <none>
+ cc = 0
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 3
+
+------------- test2.2 ----------------
+UTF8: 00
+ cc = 0
+ dst address difference: 1 dst len: 999
+ src address difference: 4 src len: 0
+UTF8: c2 80 df bf
+ cc = 0
+ dst address difference: 4 dst len: 996
+ src address difference: 8 src len: 2
+UTF8: e0 a0 80
+ cc = 0
+ dst address difference: 3 dst len: 997
+ src address difference: 4 src len: 1
+UTF8: f0 90 80 80 f4 8f bf bf f0 90 80 81 f4 8f bf be f0 92 8d 85
+ cc = 0
+ dst address difference: 20 dst len: 980
+ src address difference: 20 src len: 1
+
+------------- test3.1 ----------------
+UTF8: <none>
+ cc = 1
+ dst len: 0
+ src address difference: 0 src len: 24
+UTF8: <none>
+ cc = 1
+ dst len: 0
+ src address difference: 0 src len: 32
+UTF8: <none>
+ cc = 1
+ dst len: 1
+ src address difference: 0 src len: 32
+UTF8: <none>
+ cc = 1
+ dst len: 0
+ src address difference: 0 src len: 36
+UTF8: <none>
+ cc = 1
+ dst len: 1
+ src address difference: 0 src len: 36
+UTF8: <none>
+ cc = 1
+ dst len: 0
+ src address difference: 0 src len: 28
+UTF8: <none>
+ cc = 1
+ dst len: 1
+ src address difference: 0 src len: 28
+UTF8: <none>
+ cc = 1
+ dst len: 2
+ src address difference: 0 src len: 28
+UTF8: <none>
+ cc = 1
+ dst len: 3
+ src address difference: 0 src len: 28
+
+------------- test4 ----------------
+UTF8: 00 7f
+ cc = 0
+ dst address difference: 2 dst len: 0
+ src address difference: 8 src len: 0
+
+------------- test5 ----------------
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 1000
+ src address difference: 0 src len: 4
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 0
+ src address difference: 0 src len: 44
+UTF8: <none>
+ cc = 2
+ dst address difference: 0 dst len: 100
+ src address difference: 0 src len: 44
+
+------------- test6 ----------------
+UTF8: 00 7f 01 7e 30 5e
+ cc = 0
+ dst address difference: 6 dst len: 994
+ src address difference: 24 src len: 0
+UTF8: c2 80 df bf c2 81 df be c4 80 cc b3 d5 95 da aa
+ cc = 0
+ dst address difference: 16 dst len: 984
+ src address difference: 32 src len: 0
+UTF8: e0 a0 80 ed 9f bf ed b0 80 ef bf bf ed b0 81 ef bf be ed ba a0 ed 80 8d ee 95 95
+ cc = 0
+ dst address difference: 27 dst len: 973
+ src address difference: 36 src len: 0
+UTF8: f0 90 80 80 f4 8f bf bf f0 90 80 81 f4 8f bf be f0 92 8d 85 f0 a3 91 96 f3 be b7 8b
+ cc = 0
+ dst address difference: 28 dst len: 972
+ src address difference: 28 src len: 0
+UTF8: 78 c4 91 e1 88 b4 f1 80 90 84
+ cc = 0
+ dst address difference: 10 dst len: 990
+ src address difference: 16 src len: 0
+
+------------- test7.0 ----------------
+UTF8: 01
+ cc = 0
+ dst address difference: 1 dst len: 999
+ src address difference: 4 src len: 0
+UTF8: 01
+ cc = 0
+ dst address difference: 1 dst len: 999
+ src address difference: 4 src len: 0
+
+------------- test7.1 ----------------
+UTF8: df be
+ cc = 0
+ dst address difference: 2 dst len: 998
+ src address difference: 4 src len: 0
+UTF8: df be
+ cc = 0
+ dst address difference: 2 dst len: 998
+ src address difference: 4 src len: 0
+
+------------- test7.2 ----------------
+UTF8: ed ba a0
+ cc = 0
+ dst address difference: 3 dst len: 997
+ src address difference: 4 src len: 0
+UTF8: ed ba a0
+ cc = 0
+ dst address difference: 3 dst len: 997
+ src address difference: 4 src len: 0
+
+------------- test7.3 ----------------
+UTF8: f0 a3 91 96
+ cc = 0
+ dst address difference: 4 dst len: 996
+ src address difference: 4 src len: 0
+UTF8: f0 a3 91 96
+ cc = 0
+ dst address difference: 4 dst len: 996
+ src address difference: 4 src len: 0
diff --git a/none/tests/s390x/cu41.vgtest b/none/tests/s390x/cu41.vgtest
new file mode 100644
index 0000000..815e594
--- /dev/null
+++ b/none/tests/s390x/cu41.vgtest
@@ -0,0 +1 @@
+prog: cu41