Message ID | 20210531055019.10149-3-yamamoto@midokura.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | linux-user changes to run docker | expand |
Le 31/05/2021 à 07:50, YAMAMOTO Takashi a écrit : > It's problematic to return AT_EXECFD as it is because the user app > would close it. > This patch opens it via /proc/self/fd instead. > > Signed-off-by: YAMAMOTO Takashi <yamamoto@midokura.com> > --- > linux-user/syscall.c | 12 +++++++++++- > 1 file changed, 11 insertions(+), 1 deletion(-) > > diff --git a/linux-user/syscall.c b/linux-user/syscall.c > index a2b03ecb8b..14a63518e2 100644 > --- a/linux-user/syscall.c > +++ b/linux-user/syscall.c > @@ -8118,7 +8118,17 @@ static int do_openat(void *cpu_env, int dirfd, const char *pathname, int flags, > > if (is_proc_myself(pathname, "exe")) { > int execfd = qemu_getauxval(AT_EXECFD); > - return execfd ? execfd : safe_openat(dirfd, exec_path, flags, mode); > + if (execfd) { > + char filename[PATH_MAX]; > + int ret; > + > + snprintf(filename, sizeof(filename), "/proc/self/fd/%d", execfd); > + ret = safe_openat(dirfd, filename, flags, mode); > + if (ret != -1) { > + return ret; > + } > + } > + return safe_openat(dirfd, exec_path, flags, mode); > } > > for (fake_open = fakes; fake_open->filename; fake_open++) { > I think a dup() would be more appropriate, or explain why not. Thanks, Laurent
On Sun, Jun 20, 2021 at 11:16 PM Laurent Vivier <laurent@vivier.eu> wrote: > > Le 31/05/2021 à 07:50, YAMAMOTO Takashi a écrit : > > It's problematic to return AT_EXECFD as it is because the user app > > would close it. > > This patch opens it via /proc/self/fd instead. > > > > Signed-off-by: YAMAMOTO Takashi <yamamoto@midokura.com> > > --- > > linux-user/syscall.c | 12 +++++++++++- > > 1 file changed, 11 insertions(+), 1 deletion(-) > > > > diff --git a/linux-user/syscall.c b/linux-user/syscall.c > > index a2b03ecb8b..14a63518e2 100644 > > --- a/linux-user/syscall.c > > +++ b/linux-user/syscall.c > > @@ -8118,7 +8118,17 @@ static int do_openat(void *cpu_env, int dirfd, const char *pathname, int flags, > > > > if (is_proc_myself(pathname, "exe")) { > > int execfd = qemu_getauxval(AT_EXECFD); > > - return execfd ? execfd : safe_openat(dirfd, exec_path, flags, mode); > > + if (execfd) { > > + char filename[PATH_MAX]; > > + int ret; > > + > > + snprintf(filename, sizeof(filename), "/proc/self/fd/%d", execfd); > > + ret = safe_openat(dirfd, filename, flags, mode); > > + if (ret != -1) { > > + return ret; > > + } > > + } > > + return safe_openat(dirfd, exec_path, flags, mode); > > } > > > > for (fake_open = fakes; fake_open->filename; fake_open++) { > > > > I think a dup() would be more appropriate, or explain why not. i did this way because dup() doesn't deal with open flags. > > Thanks, > Laurent
diff --git a/linux-user/syscall.c b/linux-user/syscall.c index a2b03ecb8b..14a63518e2 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -8118,7 +8118,17 @@ static int do_openat(void *cpu_env, int dirfd, const char *pathname, int flags, if (is_proc_myself(pathname, "exe")) { int execfd = qemu_getauxval(AT_EXECFD); - return execfd ? execfd : safe_openat(dirfd, exec_path, flags, mode); + if (execfd) { + char filename[PATH_MAX]; + int ret; + + snprintf(filename, sizeof(filename), "/proc/self/fd/%d", execfd); + ret = safe_openat(dirfd, filename, flags, mode); + if (ret != -1) { + return ret; + } + } + return safe_openat(dirfd, exec_path, flags, mode); } for (fake_open = fakes; fake_open->filename; fake_open++) {
It's problematic to return AT_EXECFD as it is because the user app would close it. This patch opens it via /proc/self/fd instead. Signed-off-by: YAMAMOTO Takashi <yamamoto@midokura.com> --- linux-user/syscall.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)