Carlos Montiers
f2ee6b1759
Fix mistake. Change jb by jbe. tiny c round (INT_MAX = 0x7FFFFFFF) to a DWORD boundary and it becomes 0x80000000. Jle treats as -214783648, but Jbe treats as 214783648. Thanks to Jason Hood for explain me this.
2014-07-10 20:41:51 -04:00
Carlos Montiers
8257829623
Fix problem using alloca function, executable crashed even with a exception handler function, when try to allocate INT_MAX. Patch provided by Jason Hood in private e-mail, when I ask to him for help. He say: Feel free to pass it on to the mailing list.
2014-07-10 00:37:20 -04:00
jiang
89000c18dc
Rename:
...
68_macro_concat.c -> 68_macro_param_list_err_1.c
69_macro_concat.c -> 69_macro_param_list_err_2.c
and Remove spaces
2014-07-01 23:54:49 +08:00
Thomas Preud'homme
b31e80a43a
Specify license of lib/armeabi.c
2014-07-01 22:01:49 +08:00
jiang
a3fc543459
bug:
...
----------------------------------------------------------------------
#define hexCh(c (c >= 10 ? 'a' + c - 10 : '0' + c)
hexCh(c);
out:
jiang@jiang:~/test$ ./tcc -E c4.c
# 1 "c4.c"
(c >= 10 ? 'a' + c - 10 : '0' + c);
---------------------------------------------------------------
#define hexCh(c/3) (c >= 10 ? 'a' + c - 10 : '0' + c)
hexCh(c);
out:
jiang@jiang:~/test$ ./tcc -E c4.c
# 1 "c4.c"
/3) (c >= 10 ? 'a' + c - 10 : '0' + c);
jiang@jiang:~/test$
after patch:
# 1 "c4.c"
c4.c:1: error: may not appear in macro parameter list: "("
jiang@jiang:~/test$
jiang@jiang:~/test$ ./tcc -E c4.c
# 1 "c4.c"
c4.c:1: error: may not appear in macro parameter list: "/"
jiang@jiang:~/test$
2014-06-29 20:35:57 +08:00
David Mertens
799512388c
Revert the many un-reviewed commits starting from early April
...
Starting early April, a number of commits were pushed to the mob branch
that did not reflect the interest, wishes, or code quality of the tcc
community. This commit reverts those commits, while cherry-picking
grishka's commit for win64 linkage issues.
2014-06-24 22:42:57 -04:00
grischka
6e0a658e96
win64: try to fix linkage
...
- revert to R_X86_64_PC32 for near calls on PE
- revert to s1->section_align set to zero by default
Untested. Compared to release_0_9_26 the pe-image looks back to
normal. There are some differences in dissassembly (r10/r11 usage)
but maybe that's ok.
2014-06-24 22:09:12 -04:00
jiang
f26fdaefd8
revert vstore()
2014-06-23 01:08:54 +08:00
Thomas Coudray
1fa0fe2786
Revert "Win: Enable use "*.def + *.c" files as library instead of *.a by "-l" option"
...
This reverts commit 7a3f6d4941
2014-06-20 13:26:29 +01:00
YX Hao
7a3f6d4941
Win: Enable use "*.def + *.c" files as library instead of *.a by "-l" option
...
example: "-lshell32" will try "shell32.def + shell32.c"
Add lib type "%s/%s.c" for tcc_add_library() ('-l' option parse)
So tcc can use all files as scripts.
2014-06-16 16:00:46 +08:00
jiang
d316836008
Let init_putz one-time generation.
...
At the same time, increase the GCC style warning
---------------------------------------------------------------------------
int main()
{
int a[10] = {[5]=5};
return 0;
}
Disassembly of section .text:
0000000000000000 <main>:
0: 55 push %rbp
1: 48 89 e5 mov %rsp,%rbp
4: 48 81 ec 30 00 00 00 sub $0x30,%rsp
b: b8 05 00 00 00 mov $0x5,%eax
10: 89 45 ec mov %eax,-0x14(%rbp)
13: 48 b8 14 00 00 00 00 movabs $0x14,%rax
1a: 00 00 00
1d: 49 89 c2 mov %rax,%r10
20: b8 00 00 00 00 mov $0x0,%eax
25: 48 89 c6 mov %rax,%rsi
28: 48 8d 45 d8 lea -0x28(%rbp),%rax
2c: 48 89 c7 mov %rax,%rdi
2f: 4c 89 d2 mov %r10,%rdx
32: b8 00 00 00 00 mov $0x0,%eax
37: e8 fc ff ff ff callq 38 <main+0x38>
3c: 48 b8 10 00 00 00 00 movabs $0x10,%rax
43: 00 00 00
46: 49 89 c2 mov %rax,%r10
49: b8 00 00 00 00 mov $0x0,%eax
4e: 48 89 c6 mov %rax,%rsi
51: 48 8d 45 f0 lea -0x10(%rbp),%rax
55: 48 89 c7 mov %rax,%rdi
58: 4c 89 d2 mov %r10,%rdx
5b: b8 00 00 00 00 mov $0x0,%eax
60: e8 fc ff ff ff callq 61 <main+0x61>
65: b8 00 00 00 00 mov $0x0,%eax
6a: e9 00 00 00 00 jmpq 6f <main+0x6f>
6f: c9 leaveq
70: c3 retq
After the patch
0000000000000000 <main>:
0: 55 push %rbp
1: 48 89 e5 mov %rsp,%rbp
4: 48 81 ec 30 00 00 00 sub $0x30,%rsp
b: 48 b8 28 00 00 00 00 movabs $0x28,%rax
12: 00 00 00
15: 49 89 c2 mov %rax,%r10
18: b8 00 00 00 00 mov $0x0,%eax
1d: 48 89 c6 mov %rax,%rsi
20: 48 8d 45 d8 lea -0x28(%rbp),%rax
24: 48 89 c7 mov %rax,%rdi
27: 4c 89 d2 mov %r10,%rdx
2a: b8 00 00 00 00 mov $0x0,%eax
2f: e8 fc ff ff ff callq 30 <main+0x30>
34: b8 05 00 00 00 mov $0x5,%eax
39: 89 45 ec mov %eax,-0x14(%rbp)
3c: b8 00 00 00 00 mov $0x0,%eax
41: e9 00 00 00 00 jmpq 46 <main+0x46>
46: c9 leaveq
47: c3 retq
-----------------------------------------------------------------------------------
"c5.c"
int main()
{
// union st
struct st
{
char c;
short s;
// char cc[];
};
// union st
struct st
ss = { 1, 2, 3};
// int a = ss;
char cb[1] = {1,2,3};
return 0;
}
c5.c:12: warning: excess elements in struct initializer
c5.c:14: warning: excess elements in array initializer
c5.c:14: warning: excess elements in array initializer
2014-05-28 22:09:49 +08:00
jiang
89134dd7b0
Considering the effect of CH_EOF on line_num
2014-05-28 21:51:50 +08:00
jiang
698b16960a
Modify the 66_macro_concat_end.expect
...
I was busy recently, forget test. Thank Austin
2014-05-28 10:38:26 +08:00
jiang
bcf60562e0
The number of rows to display warning is wrong
...
For example:
#define TOK_ASM_weak TOK_WEAK1
#define TOK_ASM_weak TOK_WEAK
Output:
C8.c:3: warning: TOK_ASM_weak redefined
2014-05-23 15:46:08 +08:00
jiang
9c78da8a32
forget Delete __va_ld_reg
2014-05-21 12:50:12 +08:00
jiang
3d608d4b54
Delete a = (a > = 0)? A: -a; \
2014-05-20 15:23:55 +08:00
jiang
0199123dd7
clean '\t'
2014-05-20 15:18:59 +08:00
jiang
f8b4f59f8b
In the local use of local stack, use a global stack in the global time
2014-05-20 14:59:37 +08:00
jiang
e5e7f488e2
int main()
...
{
struct st {
int aa:16;
int bb:16;
} s;
s.aa = 1;
s.bb = 2;
return 0;
}
objdump -d:
elf64-x86-64
Disassembly of section .text:
0000000000000000 <main>:
0: 55 push %rbp
1: 48 89 e5 mov %rsp,%rbp
4: 48 81 ec 10 00 00 00 sub $0x10,%rsp
b: b8 01 00 00 00 mov $0x1,%eax
10: 48 89 c1 mov %rax,%rcx
13: 81 e1 ff ff 00 00 and $0xffff,%ecx
19: 8b 55 fc mov -0x4(%rbp),%edx
1c: 81 e2 00 00 ff ff and $0xffff0000,%edx
22: 09 d1 or %edx,%ecx
24: 89 4d fc mov %ecx,-0x4(%rbp)
27: b8 02 00 00 00 mov $0x2,%eax
2c: 48 89 c1 mov %rax,%rcx
2f: 81 e1 ff ff 00 00 and $0xffff,%ecx
35: c1 e1 10 shl $0x10,%ecx
38: 8b 55 fc mov -0x4(%rbp),%edx
3b: 81 e2 ff ff 00 00 and $0xffff,%edx
41: 09 d1 or %edx,%ecx
43: 89 4d fc mov %ecx,-0x4(%rbp)
46: b8 00 00 00 00 mov $0x0,%eax
4b: e9 00 00 00 00 jmpq 50 <main+0x50>
50: c9 leaveq
51: c3 retq
After the patch
Disassembly of section .text:
0000000000000000 <main>:
0: 55 push %rbp
1: 48 89 e5 mov %rsp,%rbp
4: 48 81 ec 10 00 00 00 sub $0x10,%rsp
b: 8b 45 fc mov -0x4(%rbp),%eax
e: 81 e0 00 00 ff ff and $0xffff0000,%eax
14: 83 c8 01 or $0x1,%eax
17: 89 45 fc mov %eax,-0x4(%rbp)
1a: 8b 45 fc mov -0x4(%rbp),%eax
1d: 81 e0 ff ff 00 00 and $0xffff,%eax
23: 81 c8 00 00 02 00 or $0x20000,%eax
29: 89 45 fc mov %eax,-0x4(%rbp)
2c: b8 00 00 00 00 mov $0x0,%eax
31: e9 00 00 00 00 jmpq 36 <main+0x36>
36: c9 leaveq
37: c3 retq
2014-05-17 12:32:00 +08:00
jiang
a94ed43094
Improved '\ n' output effect (# pragma pack (push, 8), # pragma pack (pop))
2014-05-17 12:06:18 +08:00
jiang
196c999515
1 macro_push and macro_pop work I made a mistake, no matter the definition does not define can be macro_push.
...
And the modified tcctest.c test
2, pack: in the compiler under the mode of s1->ppfp, I have no clear ideas1->ppfp
Some advice thank you Roy to me.This patch, I hope I can pass the Roy test
2014-05-17 00:55:02 +08:00
jiang
276553c6b2
add push_macro test again
2014-05-16 15:44:51 +08:00
jiang
52891b6ff6
fix push_macro, asked Tom to help me testfix push_macro
2014-05-16 12:15:00 +08:00
jiang
5a514107c4
When tcc.exe update, abitest-tcc.exe not updated. For security, you must first clean up
2014-05-15 12:09:43 +08:00
jiang
5d0785d0e1
Add warning 4
...
num en{a1,a2,a3};
enum en ee;
ee = 0xffffffffff;
char a;
a = 0xffff;
2014-05-14 21:01:00 +08:00
jiang
fcb3772a34
Add warning 3
...
struct st {int a;} ss;
int b;
b = ss;
2014-05-14 20:44:53 +08:00
jiang
07614b5e22
clean '\t'
2014-05-14 12:45:58 +08:00
jiang
c6345b5a8a
restore 2dd8587c2f32d17a2cd0443a60a614a3fa9bbe29
2014-05-13 22:05:38 +08:00
jiang
72f466c24c
Modify tcc_error ("invalid type",);
2014-05-13 22:00:42 +08:00
jiang
8d5e0cf083
fix 14d0aa450f
2014-05-13 21:57:35 +08:00
jiang
03687729ec
Add warning 2
...
For example:
struct A {
int b: 16;
int c: 16
};
sizeof (struct A);
2014-05-09 22:41:45 +08:00
jiang
14d0aa450f
Add warning
...
For example:
struct A {
int b [];
};
2014-05-09 22:35:19 +08:00
grischka
0f51ccd4e4
win64: try to fix linkage
...
- revert to R_X86_64_PC32 for near calls on PE
- revert to s1->section_align set to zero by default
Untested. Compared to release_0_9_26 the pe-image looks back to
normal. There are some differences in dissassembly (r10/r11 usage)
but maybe that's ok.
2014-05-08 17:32:29 +02:00
jiang
ad787abea6
fix bug if (seen_reg_num + reg_count <= 8)
2014-05-08 15:39:50 +08:00
grischka
899d26605c
Revert "update static void parse_number()"
...
because:
- Constructing fp numbers isn't quite trivial
- 3 additional calls to strchr per number is noticeable slow
Also: exclude abitest.c:ret_longdouble_test2 on _WIN32
for mixed gcc/tcc scenario
test case:
- make -k test (on win32):
-2.120000 0.500000 23000000000.000000
+2.120000 0.500000 22999999999.999996
...
ret_longdouble_test2... failure
This reverts 857f7dbfa6deaee6c249
2014-05-06 18:24:41 +02:00
jiang
5e56fb635a
Return to: e20c1eb99e
...
1: The new patch for the other machines still have the problem.
2: libcrt Rename (what if gcc had libcrt as well)
3: parse_number exact problem
4: VT_VLS is to allow tcc
Compile the following
int b = 9;
struct st {
int a;
int b [b]
};
struct st st1;
st1.b [8] = 9;
printf ("% d \ n", st1.b [8]);
tcc a problem. Due to problems in front, and now can not be improved
5: they commit much, bug difficult to lock, you can not let other people help develop.
6: ('\ t') too
Thanks to Michael and Ray
Their criticism I have benefited!
2014-05-04 13:18:31 +08:00
jiang
089dea355a
tcc on i386 are still having problems at work.Thank Roy report again. Struck on several variables can be connected to commit in the register. I am worried whether tcc can run the os. Since my machine is ubuntu 64 bits I can test my machine.
2014-05-03 23:51:09 +08:00
jiang
a0d45c1bcd
forget commit tccge.c for i386
2014-05-03 00:39:40 +08:00
jiang
6755b4a3de
Modify i386-gen.c,
...
Thank Roy Tam reported problems
2014-05-03 00:33:19 +08:00
jiang
e647c3137d
Fix x86-64 vla
...
For example, it should look like this
High stack
-------------
----- Func_ret_sub
------------
---- Vla stack
-------------
---- Known loc
--------------
Low
Increased loc_stack () function is used for temporary stack management, call save_reg (), released by load ()
Like this
Before use
High
----- Known loc
----
---
----
---- Pop_stack
Low
loc_stack (size, 1)
After use
High
----- Known loc
---- Pop_stack
---
----
----
Low
2014-05-02 23:45:48 +08:00
jiang
6c8207633f
Fixes include the double quotes bug
...
Added push_macro, pop_macro support
Fix pack bug, when output with-E will pack the bug
2014-05-02 11:23:54 +08:00
jiang
5b52a44b52
gen_putz () and struct_copy (), is to reduce the third-party call that
...
generates faster code tcc
Now only for x86-64
parse_number also to reduce the reliance on third-party libraries, allowing
faster analysis tcc
2014-05-02 09:42:33 +08:00
jiang
ee99fd45ab
Add a comment.
...
ref: 2a8905c93b
2014-05-02 00:39:50 +08:00
grischka
504d40a328
Attention: never use hard tabs other than 8 (eight) wide
...
The attached fix (libcrt.c) is just one example. There seem
many more such introduced from latest commits that look badly
formatted in anyone else's editors and should be fixed.
General recommended policy:
- if possible, do not add new hard tabs ('\t') at all.
Use spaces (soft tabs) instead
- in any case, configure your editor to read/write hard tabs
with width of 8 (eight)
Also:
- Avoid merge commits (unless for very good reason). Instead
use "git cherry-pick" or "git rebase" to put your commits on
top of the public branch. ref: 2a8905c93b
2014-05-01 15:33:29 +02:00
jiang
59a22d59a2
update for x86_64-gen.c
2014-05-01 20:58:43 +08:00
jiang
2742fbcf95
clean
2014-05-01 15:19:03 +08:00
jiang
87a850f553
fix its own making bug. Improved init_putz (). Modify the tests / Makefile to make the test more secure
2014-05-01 15:15:01 +08:00
jiang
9e3713facd
Expansion code again for x86_64-gen
2014-05-01 01:48:50 +08:00
jiang
2b2e7f85d7
rename i386-tok.h i386-asm.c, add PRINTF_ASM_CODE
2014-04-30 19:30:30 +08:00
jiang
ba61fd9cd1
rename libtcc1.c
2014-04-30 17:35:44 +08:00
