From: "Cheng Jin" <jincheng@ca.ibm.com>
To: Anthony Green <green@moxielogic.com>
Cc: libffi-discuss <libffi-discuss@sourceware.org>
Subject: RE: Incorrect data detected in the nested float struct with x86/libffi on Linux/64bit
Date: Wed, 9 Jun 2021 13:53:32 -0400 [thread overview]
Message-ID: <OFF6D0F33B.B956BA21-ON002586EF.00623CEB-852586EF.006248F5@ibm.com> (raw)
In-Reply-To: <CACxje58wD-98bsScOT4OZ1ZXQxLuyKAwMmO85fDWRyt1Z-7wNw@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 5628 bytes --]
Hi Anthony,
Many thanks for the quick response. The PR has been created at
https://github.com/libffi/libffi/pull/640 as requested.
Thanks and Best Regards
Cheng Jin
From: Anthony Green <green@moxielogic.com>
To: Cheng Jin <jincheng@ca.ibm.com>
Cc: libffi-discuss <libffi-discuss@sourceware.org>
Date: 2021-06-09 12:50 PM
Subject: [EXTERNAL] Re: Incorrect data detected in the nested float
struct with x86/libffi on Linux/64bit
Thank you, Cheng. Are you able to submit this test case as a github pull
request? The resulting CI testing will give us a broader picture of where
we have problems. Thanks, AG On Wed, Jun 9, 2021 at 12:41 PM Cheng Jin via
Libffi-discuss < ZjQcmQRYFpfptBannerStart
This Message Is From an External Sender
This message came from outside your organization.
ZjQcmQRYFpfptBannerEnd
Thank you, Cheng. Are you able to submit this test case as a github pull
request? The resulting CI testing will give us a broader picture of where
we have problems.
Thanks,
AG
On Wed, Jun 9, 2021 at 12:41 PM Cheng Jin via Libffi-discuss <
libffi-discuss@sourceware.org> wrote:
Hi,
We found that the nested struct [[float, float], float] works good with
x86/libffi (at https://github.com/libffi/libffi/tree/master/src/x86)
on Linux/64bit (verified on Ubuntu 19) while a variant of it like
[float,[float,float]] was messed up at the 3rd element when invoking
ffi_call() in the following test:
typedef struct stru_Nested_F stru_Nested_F;
struct stru_FF {
float elem1;
float elem2;
};
struct stru_Nested_F {
float elem1;
stru_FF elem2;
};
static float testNestedFloatStruct(float arg1, stru_Nested_F arg2)
{
float floatSum = arg1 + arg2.elem1 + arg2.elem2.elem1 +
arg2.elem2.elem2;
return floatSum;
}
int main() {
float returnValue = 0;
ffi_cif cif_temp;
UDATA structElemNum = 2;
UDATA nestedStructElemNum = 2;
UDATA argNum = 2;
ffi_type **cls_struct_fields1 = (ffi_type
**)malloc(sizeof(ffi_type
*) * (structElemNum + 1));
ffi_type **cls_struct_fields2 = (ffi_type
**)malloc(sizeof(ffi_type
*) * (nestedStructElemNum + 1));
ffi_type **dbl_arg_types = (ffi_type **)malloc(sizeof(ffi_type
*) *
(argNum + 1));
void **args_db = (void **)malloc(sizeof(void *) * (argNum +
1));
ffi_type cls_struct_type1, cls_struct_type2;
ffi_type *retType = &ffi_type_float;
float arg1;
float *arg2 = (float *)malloc(sizeof(stru_Nested_F));
cls_struct_fields2[0] = &ffi_type_float;
cls_struct_fields2[1] = &ffi_type_float;
cls_struct_fields2[2] = NULL;
cls_struct_type2.size = 0;
cls_struct_type2.alignment = 0;
cls_struct_type2.type = FFI_TYPE_STRUCT;
cls_struct_type2.elements = cls_struct_fields2;
cls_struct_fields1[0] = &ffi_type_float;
cls_struct_fields1[1] = &cls_struct_type2;
cls_struct_fields1[2] = NULL;
cls_struct_type1.size = 0;
cls_struct_type1.alignment = 0;
cls_struct_type1.type = FFI_TYPE_STRUCT;
cls_struct_type1.elements = cls_struct_fields1;
dbl_arg_types[0] = &ffi_type_float;
dbl_arg_types[1] = &cls_struct_type1;
dbl_arg_types[2] = NULL;
ffi_prep_cif(&cif_temp, FFI_DEFAULT_ABI, 2, retType,
dbl_arg_types);
arg1 = 37.88;
arg2[0] = 31.22;
arg2[1] = 33.44;
arg2[2] = 35.66;
args_db[0] = &arg1;
args_db[1] = arg2;
args_db[2] = NULL;
ffi_call(&cif_temp, FFI_FN(testNestedFloatStruct), &returnValue,
args_db);
free(cls_struct_fields1);
free(cls_struct_fields2);
free(dbl_arg_types);
free(args_db);
return 0;
}
with the debugging output as follows:
(gdb) p arg2[0]
$1 = 31.2199993
(gdb) p arg2[1]
$2 = 33.4399986
(gdb) p arg2[2]
$3 = 35.6599998
(gdb) x/10x arg2
0x7ffff025b540: 0x41f9c28f 0x4205c28f 0x420ea3d7
0xdddddddd
0x7ffff025b550: 0xb7654321 0x475d4352 0x0000000c
0x00000000
0x7ffff025b560: 0xf749c970 0x00007fff
(gdb) bt
#0 testNestedFloatStruct(arg1=37.8800011, arg2=...) at main.c:xxx
#1 0x00007ffff746e4e0 in ffi_call_unix64 () at x86/unix64.S:xxx
#2 0x00007ffff746dac8 in ffi_call (cif=0x7ffff7d44e90,
fn=0x7ffff7f9e060,
rvalue=0x16818, avalue=0x7ffff7d44f30) at x86/ffi64.c:xxx
Thread 2 "main" hit Breakpoint 2, testNestedFloatStruct(arg1=37.8800011,
arg2=...) at main.c
float floatSum = arg1 + arg2.elem1 + arg2.elem2.elem1 +
arg2.elem2.elem2;
(gdb) p arg2
$4 = {
elem1 = 31.2199993,
elem2 = {
elem1 = 33.4399986,
elem2 = 4.20389539e-45 <------------------ messed up
}
}
(gdb) x/10x &arg2
0x7ffff7d44458: 0x41f9c28f 0x4205c28f 0x00000003
0x00007fff
0x7ffff7d44468: 0xf70cf1e6 0x4217851f 0xf7d44510
0x00007fff
0x7ffff7d44478: 0xf7d44660 0x00007fff
-------------------------------------------
It looks like there is something wrong in dealing with the nested stuct
on
Linux/64bit in x86/libffi as the test above works good on Windows/64bit.
Thanks and Best Regards
Cheng Jin
[-- Attachment #2: graycol.gif --]
[-- Type: image/gif, Size: 105 bytes --]
next prev parent reply other threads:[~2021-06-09 17:53 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-09 16:41 Cheng Jin
2021-06-09 16:50 ` Anthony Green
2021-06-09 17:53 ` Cheng Jin [this message]
2021-06-15 18:02 ` Jakub Jelinek
2021-06-15 19:02 ` Jakub Jelinek
2021-06-15 19:29 ` Anthony Green
2021-06-09 23:48 ` Kaz Kylheku (libffi)
2021-06-10 0:21 ` Cheng Jin
2021-06-10 2:05 ` Kaz Kylheku (libffi)
2021-06-10 2:39 ` Cheng Jin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=OFF6D0F33B.B956BA21-ON002586EF.00623CEB-852586EF.006248F5@ibm.com \
--to=jincheng@ca.ibm.com \
--cc=green@moxielogic.com \
--cc=libffi-discuss@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).