Todd C. Miller
69fa04f069
Include the cert or ca file in error messages where applicable.
2022-05-17 10:44:18 -06:00
Todd C. Miller
a82fe9aa06
Add missing include of string.h for strerror(3).
2022-05-17 09:31:22 -06:00
Todd C. Miller
d7df7abf87
If ERR_reason_error_string() returns NULL, fall back on strerror(errno).
...
That way we get reasonable error messages for missing files, etc.
2022-05-17 09:18:03 -06:00
Todd C. Miller
3ba807b70f
set_dhparams: pass BIO_new_file() "r" for the file mode, not O_RDONLY.
...
Unlike BIO_new_fp(), BIO_new_file() takes an fopen-style mode string.
2022-05-17 08:38:49 -06:00
Todd C. Miller
161a972600
The set_sc_arg3, get_sc_arg3 and set_sc_arg4 functions are not used.
...
Use ifdef notyet to disable for now since they may be used in the
future.
2022-05-17 08:25:59 -06:00
Todd C. Miller
0bfbef0169
Use __x86_64__ preprocessor symbol, not __amd64__
...
Also clarify a comment about MIPS ptrace.
2022-05-16 16:04:24 -06:00
Todd C. Miller
7329a4a864
ptrace support has been tested on Debian/s390x.
...
It should also work on s390 but this has not been tested.
I have not added a compat mode to trace 31-bit binaries on s390x
due to the lack of a test system.
2022-05-16 13:22:31 -06:00
Todd C. Miller
2eb8ff17be
Define sudo_pt_regs instead of user_pt_regs and include the struct keyword.
...
On s390, the struct is typedef'd without a name.
2022-05-16 12:39:59 -06:00
Todd C. Miller
74c6353a84
ptrace support has been tested on Debian/riscv64.
2022-05-16 09:08:45 -06:00
Todd C. Miller
906eb19ece
Add maxseq setting to log_output example.
...
This should make it more obvious that you need to adjust maxseq
unless you have (virtually) unlimited disk space.
2022-05-15 18:46:24 -06:00
Todd C. Miller
b61a55eb1b
Fix dependency check for libssl on Debian/Ubuntu with OpenSSL 3.
...
Also add check for python 3.10 and 3.11 and remove versions < 3.4.
Fixes building on Ubuntu 22.04.
2022-05-15 07:45:51 -06:00
Todd C. Miller
55bd6272c0
Tracing 32-bit arm binaries from a 64-bit sudo works.
2022-05-14 14:41:18 -06:00
Todd C. Miller
3238c43f92
ptrace_write_string: the terminating NUL fix was reverted by mistake.
2022-05-14 13:05:00 -06:00
Todd C. Miller
cb95b75ffa
ptrace-based intercept has now been tested on 32-bit arm
2022-05-14 13:03:19 -06:00
Todd C. Miller
bb3c4ba07c
Don't use PTRACE_SET_SYSCALL for 32-bit arm binaries running on aarch64.
...
Use PTRACE_SETREGSET with NT_ARM_SYSTEM_CALL instead just like we
would for a 64-bit binary. Newer Linux headers don't define
PTRACE_SET_SYSCALL for aarch64.
2022-05-13 18:15:28 -06:00
Todd C. Miller
6a196190cf
Replace verbose flag with debug flag.
...
This is more accurate since it actually uses the debug subsystem.
2022-05-13 07:55:05 -06:00
Todd C. Miller
3226c2df40
Initial cut at MIPS support, untested.
...
Mips is a bit different in that most Linux distros appear to use
the n32 ABI on 64-bit CPUs. We don't currently support tracing a
64-bit binary from a 32-bit sudo. We could suport tracing o32
ABI binaries in compat mode, though.
2022-05-13 07:07:32 -06:00
Todd C. Miller
f4cc99c063
Add have_seccomp_action("trap") call to check for SECCOMP_MODE_FILTER.
2022-05-12 15:46:53 -06:00
Todd C. Miller
b99debf92f
Add arm-specific code to set the system call number.
...
Fixes rejection of commands due to policy on arm when in intercept mode.
2022-05-12 10:23:27 -06:00
Todd C. Miller
4a7d26c381
Fix OS major version detection on CentOS Stream
2022-05-12 09:42:34 -06:00
Todd C. Miller
ea2bf7f1f2
ptrace_intercept_execve: read back the updated syscall args in test mode.
...
This makes it easier to detect problems with the syscall rewrite code
when testing with test_ptrace.
2022-05-12 08:32:31 -06:00
Todd C. Miller
4ecada04a4
Repair ptrace_write_vec() for compat binaries.
2022-05-12 09:21:02 -06:00
Todd C. Miller
0f104d2ff1
Fix a crash when not run in verbose mode.
2022-05-12 08:37:48 -06:00
Todd C. Miller
f94e5d40db
Fix compilation error on systems with no compat arch.
...
Currently only affects i386.
2022-05-11 20:07:59 -06:00
Todd C. Miller
64aba8f88e
Move code to write a string vector to ptrace_write_vec().
2022-05-11 20:07:59 -06:00
Todd C. Miller
55f5e389ba
Fix tracing compat binaries on big endian systems.
...
We need to swap the order of the two 32-bit addresses for big-endian.
2022-05-11 20:07:59 -06:00
Todd C. Miller
31e8506c82
Enable ptrace intercept on powerpc.
...
Tested on ppc64 and ppc64le.
2022-05-11 20:07:59 -06:00
Todd C. Miller
040e75a07b
Add test_ptrace program to test ptrace-based intercept support.
2022-05-11 20:07:55 -06:00
Todd C. Miller
532e8218b2
Use unsigned long for addresses so we don't have to worry about sign extension.
2022-05-11 08:04:58 -06:00
Todd C. Miller
6cd461d9f8
ptrace_write_string: make sure we always write the terminating NUL.
...
We can't check *str for NUL since it may not have been written yet.
2022-05-10 11:23:44 -06:00
Todd C. Miller
8fe1238679
Fix compilation error when SECCOMP_AUDIT_ARCH_COMPAT is not defined.
2022-05-10 08:17:00 -06:00
Todd C. Miller
a40d62b3d5
It is now safe to make WORDALIGN use compat (not native) aligment.
...
We allocate space for an extra pointer between argv and the string
table for compat binaries so there is no need to align address to
sizeof(long).
2022-05-09 11:03:30 -06:00
Todd C. Miller
f82aa334cc
Use the entire word in ptrace_get_vec_len() and ptrace_read_vec().
...
For compat binaries, use the upper 32-bits as the next word instead
of calling ptrace(2) to get it. This reduces the number of ptrace(2)
calls when reading argv and envp for compat binaries.
2022-05-09 10:09:30 -06:00
Todd C. Miller
c9a481506d
We don't need to align strings in the string table.
...
We align the start of the string table to a word boundary to help
prevent overlap when writing the pointers. However, the actual
strings themselves don't need to be aligned.
2022-05-07 18:42:32 -06:00
Todd C. Miller
206cd10ed5
Avoid potentially overwriting string table when writing argv.
...
In compat mode, if argc is odd, writing the last pointer of argv will
overlap with the address of argv[0], so leave an extra word in between.
Also remove incorrect comments about PTRACE_PEEKDATA unaligned access.
2022-05-06 19:46:22 -06:00
Todd C. Miller
0a2975367e
ptrace_intercept_execve: rewrite path to exec if changed by the policy
2022-05-05 19:07:12 -06:00
Todd C. Miller
a15ca9c85b
Use native word size for padding and when reading/writing strings.
...
If we try to use the compat word size we can end up in a situation
where a subsequent PTRACE_POKEDATA overwrites part of what we've
already written since it always writes in sizeof(long) units.
2022-05-06 10:46:06 -06:00
Todd C. Miller
339746730c
ptrace_intercept_execve: plug memory leak of get_execve_info() buffer
2022-05-05 19:06:19 -06:00
Todd C. Miller
35ea534b3e
Move register definitions to exec_ptrace.h
2022-05-05 13:37:26 -06:00
Todd C. Miller
d3a1bf4216
Add support for intercepting 32-bit binaries on 64-bit systems.
...
We need to define the ptrace register struct ourselves for the
32-bit system since there is no good way to get it from the system
headers. Currently only implemented for x86_64 and aarch64.
2022-05-05 09:29:05 -06:00
Todd C. Miller
de678ba775
Add setters and getters for ptrace(2) register access.
...
This will be used when running 32-bit binaries from a 64-bit sudo.
2022-05-05 09:17:58 -06:00
Todd C. Miller
f53053a66f
exec_ptrace_handled: don't return early if ptrace_intercept_execve() fails.
...
We need to continue the traced process even if there is a fatal
error. Otherwise, sudo will appear to hang as the running process
is left in PTRACE_EVENT stop.
2022-05-05 09:15:50 -06:00
Todd C. Miller
bbbb6e2ace
Don't use PTRACE_GETREGS, it is too complicated when runing compat binaries.
...
Unlike PTRACE_GETREGSET, PTRACE_GETREGS requires that we manually
map registers from 64-bit to 32-bit layouts when running, e.g. a
32-bit binary from a 64-bit sudo process.
2022-05-05 08:53:51 -06:00
Todd C. Miller
4ab6a87b96
Initialize intercept_allow_setid to true if we use ptrace(2) and seccomp(2).
2022-05-04 13:32:28 -06:00
Todd C. Miller
e84fdd99fd
If the process is already being traced, just resume it and clear flags.
...
This makes it possible to run sudo in ptrace intercept mode from within
a shell (or other process) that is already being traced by sudo.
2022-05-03 13:34:40 -06:00
Todd C. Miller
cc52ab770c
exec_ptrace_handled: fix delivery of non-stop signals.
...
We need to deliver signals to the tracee as long as it is not
a group stop. Fixes a hang while tracing another sudo process.
2022-05-03 12:54:10 -06:00
Todd C. Miller
4cac34b86d
Make SIGCHLD handler more consistent with the pty version.
...
No real change other than a few debug statements.
2022-05-03 09:38:28 -06:00
Todd C. Miller
4d75b32799
sudoers_lookup_check: preserve intercepted flag when reinitializing cmnd_info
...
Otherwise we may not reject an attempt to run a set-user-ID command.
2022-05-03 09:30:09 -06:00
Todd C. Miller
3ee8bcefb0
Kill the command if intercept_setup() or ptrace_seize() fail.
2022-05-03 09:25:58 -06:00
Todd C. Miller
1d17415b69
Add support for intercepting x32 binaries on Linux x64_64.
2022-05-02 14:36:34 -06:00