2022-01-24 21:46:15 +08:00
|
|
|
/* Check for expected la_objopen and la_objeclose for all objects.
|
2023-01-07 05:08:04 +08:00
|
|
|
Copyright (C) 2022-2023 Free Software Foundation, Inc.
|
2022-01-24 21:46:15 +08:00
|
|
|
This file is part of the GNU C Library.
|
|
|
|
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
|
|
|
modify it under the terms of the GNU Lesser General Public
|
|
|
|
License as published by the Free Software Foundation; either
|
|
|
|
version 2.1 of the License, or (at your option) any later version.
|
|
|
|
|
|
|
|
The GNU C Library is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
Lesser General Public License for more details.
|
|
|
|
|
|
|
|
You should have received a copy of the GNU Lesser General Public
|
|
|
|
License along with the GNU C Library; if not, see
|
|
|
|
<https://www.gnu.org/licenses/>. */
|
|
|
|
|
|
|
|
#include <array_length.h>
|
|
|
|
#include <errno.h>
|
|
|
|
#include <getopt.h>
|
|
|
|
#include <link.h>
|
|
|
|
#include <limits.h>
|
|
|
|
#include <inttypes.h>
|
|
|
|
#include <gnu/lib-names.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <support/capture_subprocess.h>
|
|
|
|
#include <support/check.h>
|
|
|
|
#include <support/xstdio.h>
|
|
|
|
#include <support/xdlfcn.h>
|
|
|
|
#include <support/support.h>
|
|
|
|
|
|
|
|
static int restart;
|
|
|
|
#define CMDLINE_OPTIONS \
|
|
|
|
{ "restart", no_argument, &restart, 1 },
|
|
|
|
|
|
|
|
static int
|
|
|
|
handle_restart (void)
|
|
|
|
{
|
|
|
|
xdlopen ("tst-audit23mod.so", RTLD_NOW);
|
|
|
|
xdlmopen (LM_ID_NEWLM, LIBC_SO, RTLD_NOW);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool
|
|
|
|
startswith (const char *str, const char *pre)
|
|
|
|
{
|
|
|
|
size_t lenpre = strlen (pre);
|
|
|
|
size_t lenstr = strlen (str);
|
|
|
|
return lenstr >= lenpre && memcmp (pre, str, lenpre) == 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool
|
|
|
|
is_vdso (const char *str)
|
|
|
|
{
|
|
|
|
return startswith (str, "linux-gate")
|
|
|
|
|| startswith (str, "linux-vdso");
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
do_test (int argc, char *argv[])
|
|
|
|
{
|
|
|
|
/* We must have either:
|
|
|
|
- One or four parameters left if called initially:
|
|
|
|
+ path to ld.so optional
|
|
|
|
+ "--library-path" optional
|
|
|
|
+ the library path optional
|
|
|
|
+ the application name */
|
|
|
|
if (restart)
|
|
|
|
return handle_restart ();
|
|
|
|
|
|
|
|
char *spargv[9];
|
|
|
|
TEST_VERIFY_EXIT (((argc - 1) + 3) < array_length (spargv));
|
|
|
|
int i = 0;
|
|
|
|
for (; i < argc - 1; i++)
|
|
|
|
spargv[i] = argv[i + 1];
|
|
|
|
spargv[i++] = (char *) "--direct";
|
|
|
|
spargv[i++] = (char *) "--restart";
|
|
|
|
spargv[i] = NULL;
|
|
|
|
|
|
|
|
setenv ("LD_AUDIT", "tst-auditmod23.so", 0);
|
|
|
|
struct support_capture_subprocess result
|
|
|
|
= support_capture_subprogram (spargv[0], spargv);
|
|
|
|
support_capture_subprocess_check (&result, "tst-audit22", 0, sc_allow_stderr);
|
|
|
|
|
|
|
|
/* The expected la_objopen/la_objclose:
|
|
|
|
1. executable
|
|
|
|
2. loader
|
|
|
|
3. libc.so
|
|
|
|
4. tst-audit23mod.so
|
|
|
|
5. libc.so (LM_ID_NEWLM).
|
|
|
|
6. vdso (optional and ignored). */
|
|
|
|
enum { max_objs = 6 };
|
|
|
|
struct la_obj_t
|
|
|
|
{
|
|
|
|
char *lname;
|
|
|
|
uintptr_t laddr;
|
|
|
|
Lmid_t lmid;
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
uintptr_t cookie;
|
|
|
|
uintptr_t namespace;
|
2022-01-24 21:46:15 +08:00
|
|
|
bool closed;
|
|
|
|
} objs[max_objs] = { [0 ... max_objs-1] = { .closed = false } };
|
|
|
|
size_t nobjs = 0;
|
|
|
|
|
|
|
|
/* The expected namespaces are one for the audit module, one for the
|
|
|
|
application, and another for the dlmopen on handle_restart. */
|
|
|
|
enum { max_ns = 3 };
|
|
|
|
uintptr_t acts[max_ns] = { 0 };
|
|
|
|
size_t nacts = 0;
|
|
|
|
int last_act = -1;
|
|
|
|
uintptr_t last_act_cookie = -1;
|
|
|
|
bool seen_first_objclose = false;
|
|
|
|
|
|
|
|
FILE *out = fmemopen (result.err.buffer, result.err.length, "r");
|
|
|
|
TEST_VERIFY (out != NULL);
|
|
|
|
char *buffer = NULL;
|
|
|
|
size_t buffer_length = 0;
|
|
|
|
while (xgetline (&buffer, &buffer_length, out))
|
|
|
|
{
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
*strchrnul (buffer, '\n') = '\0';
|
|
|
|
printf ("info: subprocess output: %s\n", buffer);
|
|
|
|
|
2022-01-24 21:46:15 +08:00
|
|
|
if (startswith (buffer, "la_activity: "))
|
|
|
|
{
|
|
|
|
uintptr_t cookie;
|
|
|
|
int this_act;
|
|
|
|
int r = sscanf (buffer, "la_activity: %d %"SCNxPTR"", &this_act,
|
|
|
|
&cookie);
|
|
|
|
TEST_COMPARE (r, 2);
|
|
|
|
|
|
|
|
if (this_act == LA_ACT_ADD && acts[nacts] != cookie)
|
|
|
|
{
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
/* The cookie identifies the object at the head of the
|
|
|
|
link map, so we only add a new namespace if it
|
|
|
|
changes from the previous one. This works since
|
|
|
|
dlmopen is the last in the test body. */
|
|
|
|
if (cookie != last_act_cookie && last_act_cookie != -1)
|
|
|
|
TEST_COMPARE (last_act, LA_ACT_CONSISTENT);
|
|
|
|
|
2022-01-24 21:46:15 +08:00
|
|
|
acts[nacts++] = cookie;
|
|
|
|
last_act_cookie = cookie;
|
|
|
|
}
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
/* LA_ACT_DELETE is called multiple times for each
|
|
|
|
namespace, depending on destruction order. */
|
2022-01-24 21:46:15 +08:00
|
|
|
else if (this_act == LA_ACT_DELETE)
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
last_act_cookie = cookie;
|
2022-01-24 21:46:15 +08:00
|
|
|
else if (this_act == LA_ACT_CONSISTENT)
|
|
|
|
{
|
|
|
|
TEST_COMPARE (cookie, last_act_cookie);
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
last_act_cookie = -1;
|
2022-01-24 21:46:15 +08:00
|
|
|
|
|
|
|
/* LA_ACT_DELETE must always be followed by an la_objclose. */
|
|
|
|
if (last_act == LA_ACT_DELETE)
|
|
|
|
TEST_COMPARE (seen_first_objclose, true);
|
|
|
|
else
|
|
|
|
TEST_COMPARE (last_act, LA_ACT_ADD);
|
|
|
|
}
|
|
|
|
|
|
|
|
last_act = this_act;
|
|
|
|
seen_first_objclose = false;
|
|
|
|
}
|
|
|
|
else if (startswith (buffer, "la_objopen: "))
|
|
|
|
{
|
|
|
|
char *lname;
|
|
|
|
uintptr_t laddr;
|
|
|
|
Lmid_t lmid;
|
|
|
|
uintptr_t cookie;
|
|
|
|
int r = sscanf (buffer, "la_objopen: %"SCNxPTR" %ms %"SCNxPTR" %ld",
|
|
|
|
&cookie, &lname, &laddr, &lmid);
|
|
|
|
TEST_COMPARE (r, 4);
|
|
|
|
|
|
|
|
/* la_objclose is not triggered by vDSO because glibc does not
|
|
|
|
unload it. */
|
|
|
|
if (is_vdso (lname))
|
|
|
|
continue;
|
|
|
|
if (nobjs == max_objs)
|
|
|
|
FAIL_EXIT1 ("non expected la_objopen: %s %"PRIxPTR" %ld",
|
|
|
|
lname, laddr, lmid);
|
|
|
|
objs[nobjs].lname = lname;
|
|
|
|
objs[nobjs].laddr = laddr;
|
|
|
|
objs[nobjs].lmid = lmid;
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
objs[nobjs].cookie = cookie;
|
|
|
|
objs[nobjs].namespace = last_act_cookie;
|
2022-01-24 21:46:15 +08:00
|
|
|
objs[nobjs].closed = false;
|
|
|
|
nobjs++;
|
|
|
|
|
|
|
|
/* This indirectly checks that la_objopen always comes before
|
|
|
|
la_objclose btween la_activity calls. */
|
|
|
|
seen_first_objclose = false;
|
|
|
|
}
|
|
|
|
else if (startswith (buffer, "la_objclose: "))
|
|
|
|
{
|
|
|
|
char *lname;
|
|
|
|
uintptr_t laddr;
|
|
|
|
Lmid_t lmid;
|
|
|
|
uintptr_t cookie;
|
|
|
|
int r = sscanf (buffer, "la_objclose: %"SCNxPTR" %ms %"SCNxPTR" %ld",
|
|
|
|
&cookie, &lname, &laddr, &lmid);
|
|
|
|
TEST_COMPARE (r, 4);
|
|
|
|
|
|
|
|
for (size_t i = 0; i < nobjs; i++)
|
|
|
|
{
|
|
|
|
if (strcmp (lname, objs[i].lname) == 0 && lmid == objs[i].lmid)
|
|
|
|
{
|
|
|
|
TEST_COMPARE (objs[i].closed, false);
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
TEST_COMPARE (objs[i].cookie, cookie);
|
|
|
|
if (objs[i].namespace == -1)
|
|
|
|
/* No LA_ACT_ADD before the first la_objopen call. */
|
|
|
|
TEST_COMPARE (acts[0], last_act_cookie);
|
|
|
|
else
|
|
|
|
TEST_COMPARE (objs[i].namespace, last_act_cookie);
|
2022-01-24 21:46:15 +08:00
|
|
|
objs[i].closed = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* la_objclose should be called after la_activity(LA_ACT_DELETE) for
|
|
|
|
the closed object's namespace. */
|
|
|
|
TEST_COMPARE (last_act, LA_ACT_DELETE);
|
elf: Always call destructors in reverse constructor order (bug 30785)
The current implementation of dlclose (and process exit) re-sorts the
link maps before calling ELF destructors. Destructor order is not the
reverse of the constructor order as a result: The second sort takes
relocation dependencies into account, and other differences can result
from ambiguous inputs, such as cycles. (The force_first handling in
_dl_sort_maps is not effective for dlclose.) After the changes in
this commit, there is still a required difference due to
dlopen/dlclose ordering by the application, but the previous
discrepancies went beyond that.
A new global (namespace-spanning) list of link maps,
_dl_init_called_list, is updated right before ELF constructors are
called from _dl_init.
In dl_close_worker, the maps variable, an on-stack variable length
array, is eliminated. (VLAs are problematic, and dlclose should not
call malloc because it cannot readily deal with malloc failure.)
Marking still-used objects uses the namespace list directly, with
next and next_idx replacing the done_index variable.
After marking, _dl_init_called_list is used to call the destructors
of now-unused maps in reverse destructor order. These destructors
can call dlopen. Previously, new objects do not have l_map_used set.
This had to change: There is no copy of the link map list anymore,
so processing would cover newly opened (and unmarked) mappings,
unloading them. Now, _dl_init (indirectly) sets l_map_used, too.
(dlclose is handled by the existing reentrancy guard.)
After _dl_init_called_list traversal, two more loops follow. The
processing order changes to the original link map order in the
namespace. Previously, dependency order was used. The difference
should not matter because relocation dependencies could already
reorder link maps in the old code.
The changes to _dl_fini remove the sorting step and replace it with
a traversal of _dl_init_called_list. The l_direct_opencount
decrement outside the loader lock is removed because it appears
incorrect: the counter manipulation could race with other dynamic
loader operations.
tst-audit23 needs adjustments to the changes in LA_ACT_DELETE
notifications. The new approach for checking la_activity should
make it clearer that la_activty calls come in pairs around namespace
updates.
The dependency sorting test cases need updates because the destructor
order is always the opposite order of constructor order, even with
relocation dependencies or cycles present.
There is a future cleanup opportunity to remove the now-constant
force_first and for_fini arguments from the _dl_sort_maps function.
Fixes commit 1df71d32fe5f5905ffd5d100e5e9ca8ad62 ("elf: Implement
force_first handling in _dl_sort_maps_dfs (bug 28937)").
Reviewed-by: DJ Delorie <dj@redhat.com>
2023-09-08 18:32:14 +08:00
|
|
|
seen_first_objclose = true;
|
2022-01-24 21:46:15 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for (size_t i = 0; i < nobjs; i++)
|
|
|
|
{
|
|
|
|
TEST_COMPARE (objs[i].closed, true);
|
|
|
|
free (objs[i].lname);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* la_activity(LA_ACT_CONSISTENT) should be the last callback received.
|
|
|
|
Since only one link map may be not-CONSISTENT at a time, this also
|
|
|
|
ensures la_activity(LA_ACT_CONSISTENT) is the last callback received
|
|
|
|
for every namespace. */
|
|
|
|
TEST_COMPARE (last_act, LA_ACT_CONSISTENT);
|
|
|
|
|
|
|
|
free (buffer);
|
|
|
|
xfclose (out);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define TEST_FUNCTION_ARGV do_test
|
|
|
|
#include <support/test-driver.c>
|