I confirm the patch fixes the deadlock I was seeing here.
--
Fabio Erculiani
TaskState contains linux_bprm struct which encapsulates argv among
other things.
argv might be used around the code and is expected to contain valid
data. Before this patch, ts->bprm->argv was NULL due to it being
freed right after loader_exec().
Signed-off-by: Fabio Erculiani
---
linu
-off-by: Fabio Erculiani
---
linux-user/syscall.c | 19 +++
1 files changed, 15 insertions(+), 4 deletions(-)
diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index 9ba51bf..3e8e3dd 100644
--- a/linux-user/syscall.c
+++ b/linux-user/syscall.c
@@ -4678,11 +4678,22 @@ static
Done, it all works now ;-) !
--
Fabio Erculiani
---
linux-user/main.c |5 -
1 files changed, 0 insertions(+), 5 deletions(-)
diff --git a/linux-user/main.c b/linux-user/main.c
index 788ff98..513d583 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -3492,11 +3492,6 @@ int main(int argc, char **argv, char **envp)
_exit(
-off-by: Fabio Erculiani
---
linux-user/syscall.c | 21 -
1 files changed, 16 insertions(+), 5 deletions(-)
diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index 9ba51bf..e563c00 100644
--- a/linux-user/syscall.c
+++ b/linux-user/syscall.c
@@ -4678,14 +4678,25
d to do it or the
TaskState lifecycle matches the executable (so there is no need to
free() it) ?
--
Fabio Erculiani
Yeah, debugging.
Moreover we have this scenario:
$ /bin/cat /proc/self/stat
32297 (cat) ..
I guess we should use basename() anyway...?
--
Fabio Erculiani
Mumble,
that is what happens already...
Let me see why I get NULL here...
--
Fabio Erculiani
How about setting ts->bprm->argv = target_argv; ?
I'm not a qemu codebase expert, but if it's always NULL (why is it
NULL?) or can be NULL...
It looks like can be done easily from main.c... without making a
variable global.
--
Fabio Erculiani
Or just using linux_binprm->filename with basename()
--
Fabio Erculiani
On Tue, Jan 3, 2012 at 6:41 PM, Alexander Graf wrote:
>
> On 03.01.2012, at 17:07, Fabio Erculiani wrote:
>
>> ts->bprm->argv seems NULL here.
>> Isn't it supposed to be set?
>
> Good question. Maybe we need some other way to fetch argv0 then?
or we co
This fixes -Wformat warning introduced by
983da8bec44fd1ab5e9730f01e719b3849aa0a35
Signed-off-by: Fabio Erculiani
---
linux-user/syscall.c |3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index f2af5d5..e563c00 100644
--- a
ts->bprm->argv seems NULL here.
Isn't it supposed to be set?
--
Fabio Erculiani
-off-by: Fabio Erculiani
---
linux-user/syscall.c | 20 +++-
1 files changed, 15 insertions(+), 5 deletions(-)
diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index 9ba51bf..f2af5d5 100644
--- a/linux-user/syscall.c
+++ b/linux-user/syscall.c
@@ -4678,14 +4678,24
hints.
>
> And thanks a lot for fixing this!
whoops, yeah this is my first contribution to qemu ;)
>
> On 03.01.2012, at 08:25, Fabio Erculiani wrote:
>
>> Hi all,
>> this is a patch on top of "[PATCH 4/5] linux-user: fake
>> /proc/self/stat" (sorry I co
d64 " ", val);
+ }
+ else {
+ /* for the rest, write zeros */
+ snprintf(buf, sizeof(buf), "0%c", i == 43 ? '\n' : ' ');
}
- snprintf(buf, sizeof(buf), "%"PRId64 "%c", val, i == 43 ? '\n' : ' ');
len = strlen(buf);
if (write(fd, buf, len) != len) {
return -1;
--
Fabio Erculiani
17 matches
Mail list logo
