qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Gavin Shan <gshan@redhat.com>
To: Richard Henderson <richard.henderson@linaro.org>,
	Paolo Bonzini <pbonzini@redhat.com>
Cc: Aldy Hernandez <aldyh@redhat.com>, Thomas Huth <thuth@redhat.com>,
	qemu-devel <qemu-devel@nongnu.org>,
	Laurent Vivier <laurent@vivier.eu>,
	qemu-arm <qemu-arm@nongnu.org>, Shan Gavin <shan.gavin@gmail.com>,
	Philippe Mathieu Daude <philmd@redhat.com>
Subject: Re: [PATCH] configure: Improve alias attribute check
Date: Mon, 22 Mar 2021 21:54:36 +1100	[thread overview]
Message-ID: <4231d15c-493f-cb3d-39ef-aff40e3618ee@redhat.com> (raw)
In-Reply-To: <0cad2de8-1150-be7c-237b-9f3ddea52473@linaro.org>

Hi Richard and Paolo,

On 3/22/21 5:23 AM, Richard Henderson wrote:
> On 3/21/21 11:46 AM, Paolo Bonzini wrote:
>> HRM, what about biting the bullet and making exec-vary.c a C++ source?... Then instead of making it conditional an attribute((alias)), we make it conditional on having a C++ compiler.
> 
> Doesn't help.  The gcc bug I filed talks about c++, because that's the closest analogy.
> 
> But set_preferred_target_page_bits is called *much* later than a constructor. Though still before any use of the variable in question, for which we have an --enable-debug-tcg assertion.
> 

It looks this issue can be avoided after "volatile" is applied to
@target_page. However, I'm not sure if it's the correct fix to have.
If it is, I can post a formal patch so that it can be included.

--- a/exec-vary.c
+++ b/exec-vary.c
-extern const TargetPageBits target_page
+extern const TargetPageBits volatile target_page
      __attribute__((alias("init_target_page")));
--- a/include/exec/cpu-all.h
+++ b/include/exec/cpu-all.h
-extern const TargetPageBits target_page;
+extern const TargetPageBits volatile target_page;


According to the experiments I did, it would be function call
to set_preferred_target_page_bits() is dropped when the machine
is created. The following c files are used in the experiment:

--- a.c ---

static int x;
const extern int VOLATILE y __attribute__((alias("x")));
extern int read_y(void);

void write_x(int val) { x = 1; }
int main(void) { return read_y(); }

--- b.c---

extern const int VOLATILE y;
extern void write_x(int val);

int read_y(void) { write_x(1); return y; }

# gcc a.c b.c -O2 -flto=auto -DVOLATILE= -o a; ./a; echo $?
0
# gdb -nw ./a
(gdb) disassem main
Dump of assembler code for function main:
    0x0000000000400480 <+0>:	adrp	x1, 0x420000 <__libc_start_main@got.plt>
    0x0000000000400484 <+4>:	mov	w2, #0x1                   	// #1
    0x0000000000400488 <+8>:	mov	w0, #0x0                   	// #0
    0x000000000040048c <+12>:	str	w2, [x1, #32]
    0x0000000000400490 <+16>:	ret
End of assembler dump.

# gcc a.c b.c -O2 -flto=auto -DVOLATILE=volatile -o a; ./a; echo $?
1
# gdb -nw ./a
(gdb) disassem main
Dump of assembler code for function main:
    0x0000000000400480 <+0>:	adrp	x1, 0x420000 <__libc_start_main@got.plt>
    0x0000000000400484 <+4>:	mov	w2, #0x1                   	// #1
    0x0000000000400488 <+8>:	adrp	x0, 0x420000 <__libc_start_main@got.plt>
    0x000000000040048c <+12>:	str	w2, [x1, #32]
    0x0000000000400490 <+16>:	ldr	w0, [x0, #32]
    0x0000000000400494 <+20>:	ret


Thanks,
Gavin



  reply	other threads:[~2021-03-22  7:57 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-20  4:27 [PATCH] configure: Improve alias attribute check Gavin Shan
2021-03-20  4:48 ` Thomas Huth
2021-03-20 23:32   ` Gavin Shan
2021-03-20 17:52 ` Paolo Bonzini
2021-03-20 22:33   ` Richard Henderson
2021-03-20 23:36     ` Gavin Shan
2021-03-21 15:49     ` Richard Henderson
2021-03-21 16:50       ` Paolo Bonzini
2021-03-21 17:34         ` Richard Henderson
2021-03-21 17:43           ` Paolo Bonzini
2021-03-21 17:46             ` Paolo Bonzini
2021-03-21 18:23               ` Richard Henderson
2021-03-22 10:54                 ` Gavin Shan [this message]
2021-03-22 20:59                   ` Richard Henderson
2021-03-23  3:13                     ` Gavin Shan

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=4231d15c-493f-cb3d-39ef-aff40e3618ee@redhat.com \
    --to=gshan@redhat.com \
    --cc=aldyh@redhat.com \
    --cc=laurent@vivier.eu \
    --cc=pbonzini@redhat.com \
    --cc=philmd@redhat.com \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=shan.gavin@gmail.com \
    --cc=thuth@redhat.com \
    /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 NNTP newsgroup(s).