RFE: Update syscalls.csv to kernel version v7.0-rc7

CONTRIBUTING.md

@@ -64,6 +64,105 @@ base, and can be enabled via the "--enable-code-coverage" configure flag and
 the "check-code-coverage" make target.  Additional details on generating code
 coverage information can be found in the .travis.yml file.
 
+## How to Update the syscalls.csv Table
+
+*** NOTE - This currently can only be done on Ubuntu ***
+
+1. Install dependencies
+
+   In addition to the normal libseccomp package dependencies, the following
+   packages must also be installed:
+   ```
+   apt install libc6-dev-x32
+   ```
+
+1. Download source packages
+
+   Download the following source packages:
+   ```
+   git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
+   git clone https://github.com/hrw/syscalls-table.git
+   git clone git@github.com:<yourrepo>/libseccomp.git
+   ```
+
+1. Add new kernel version enumerations
+
+   The first line of [src/syscalls.csv](https://github.com/seccomp/libseccomp/blob/main/src/syscalls.csv)
+   contains the newest kernel version known by libseccomp.  Add new kernel
+   version enumerations to the end of the `enum scmp_kver` enumeration in
+   [seccomp-kvers.h](https://github.com/seccomp/libseccomp/blob/main/include/seccomp-kvers.h).
+
+   Optional - Add new kernel versions to the `kernel_versions` list in
+   [arch-build-kver-tables.py](https://github.com/seccomp/libseccomp/blob/main/src/arch-build-kver-tables.py).
+
+1. Build the table(s) of architectures, syscalls, and syscall numbers
+
+   Using the [syscalls-table](https://github.com/hrw/syscalls-table) tool,
+   build the tables of architectures, syscalls, and syscall numbers for the
+   new kernel versions.
+
+   ```
+   cd libseccomp
+   ./src/arch-build-kver-tables.py -d ../syscalls-table -k ../linux -V [Kernel Version(s)]
+
+   # example:
+   ./src/arch-build-kver-tables.py -d ../syscalls-table -k ../linux -V 6.14,6.15,6.16,6.17,6.18,6.19,7.0-rc7
+   ```
+
+1. Add the tables to syscalls.csv
+
+   Parse the tables generated in the previous step and add the data to
+   syscalls.csv.
+
+   ```
+   ./src/arch-update-syscalls-csv.py -a -d ./ -k ../linux -c src/syscalls.csv -V [Kernel Version(s)]
+
+   # example
+   ./src/arch-update-syscalls-csv.py -a -d ./ -k ../linux -c src/syscalls.csv -V 6.14,6.15,6.16,6.17,6.18,6.19,7.0-rc7
+   ```
+
+1. Update seccomp-syscalls.h with new syscalls
+
+   Run `cd src && ./arch-syscall-check` to determine if any new syscalls were
+   added and if they require __PNR and/or __SNR definitions.  If this tool
+   identifies missing definitions, add them to
+   [include/seccomp-syscalls.h](https://github.com/seccomp/libseccomp/blob/main/include/seccomp-syscalls.h).  [Here](https://github.com/seccomp/libseccomp/commit/f01e67509e45c672f4bdd643d94d90867cc19d90)
+    is an example of the syscalls that were added to kernel version v6.12.
+
+
+1. Build the legacy syscalls.csv table (optional but recommended)
+
+   Prior to tracking the kernel version where syscalls were added, libseccomp
+   employed internal tools to build the syscalls.csv table.  These tools can
+   be used to validate the syscall numbers and their architectures.  Note that
+   they cannot be used to validate the kernel version number.
+
+   ```
+   ./autogen.sh && ./configure --enable-python && make check-build
+
+   cd src
+   make arch-syscall-dump
+   ./arch-syscall-validate -c syscalls-prev.csv ../../linux/
+   ```
+
+1. Compare CSVs
+
+   Compare the checked-in (HEAD) CSV with the newly-generated syscalls.csv.
+   Verify the following:
+   * All new syscall names were properly added
+   * If a syscall number changed, it should only have transitioned from `PNR`
+     to a valid number.  If a number changed for an architecture, verify that
+     its associated kernel version is correct
+   * No syscall rows were deleted
+
+   If you built `syscalls-prev.csv` in the previous step, do the same comparisons
+   as outlined above.  Again, note that `syscalls-prev.csv` does not contain
+   kernel version information, so only the syscall names, syscall numbers, and
+   architectures can be verified.
+
+   There are many tools to compare CSVs.  This [tool](https://www.textcompare.org/csv/)
+   has been especially useful.
+
 ## Explain Your Work
 
 At the top of every patch you should include a description of the problem you

include/seccomp-kvers.h

@@ -108,5 +108,8 @@ enum scmp_kver {
 	SCMP_KV_6_15 = 78,
 	SCMP_KV_6_16 = 79,
 	SCMP_KV_6_17 = 80,
+	SCMP_KV_6_18 = 81,
+	SCMP_KV_6_19 = 82,
+	SCMP_KV_7_0 = 83,
 	__SCMP_KV_MAX,
 };

include/seccomp-syscalls.h

@@ -283,6 +283,9 @@
 #define __PNR_getpagesize			-10249
 #define __PNR_riscv_hwprobe			-10250
 #define __PNR_uretprobe				-10251
+#define __PNR_uprobe				-10252
+#define __PNR_rseq_slice_yield			-10253
+#define __PNR_listns				-10254
 
 /*
  * libseccomp syscall definitions
@@ -648,6 +651,9 @@
 
 #define __SNR_fgetxattr			__NR_fgetxattr
 
+#define __SNR_file_getattr		__NR_file_getattr
+#define __SNR_file_setattr		__NR_file_setattr
+
 #define __SNR_finit_module		__NR_finit_module
 
 #define __SNR_flistxattr		__NR_flistxattr
@@ -1082,6 +1088,8 @@
 
 #define __SNR_listmount			__NR_listmount
 
+#define __SNR_listns			__NR_listns
+
 #define __SNR_listxattr			__NR_listxattr
 
 #define __SNR_listxattrat		__NR_listxattrat
@@ -1356,6 +1364,8 @@
 #define __SNR_open_tree			__PNR_open_tree
 #endif
 
+#define __SNR_open_tree_attr		__NR_open_tree_attr
+
 #define __SNR_openat			__NR_openat
 
 #define __SNR_openat2			__NR_openat2
@@ -1614,6 +1624,8 @@
 #define __SNR_rseq			__PNR_rseq
 #endif
 
+#define __SNR_rseq_slice_yield		__NR_rseq_slice_yield
+
 #define __SNR_rt_sigaction		__NR_rt_sigaction
 
 #define __SNR_rt_sigpending		__NR_rt_sigpending
@@ -2304,6 +2316,12 @@
 
 #define __SNR_unshare			__NR_unshare
 
+#ifdef __NR_uprobe
+#define __SNR_uprobe			__NR_uprobe
+#else
+#define __SNR_uprobe			__PNR_uprobe
+#endif
+
 #ifdef __NR_uretprobe
 #define __SNR_uretprobe			__NR_uretprobe
 #else

src/arch-build-kver-tables.py

@@ -29,6 +29,7 @@
 #######################################################
 
 from subprocess import TimeoutExpired
+from pathlib import Path
 import subprocess
 import argparse
 import os
@@ -43,7 +44,7 @@
                    '5.11', '5.12', '5.13', '5.14', '5.15', '5.16', '5.17',
                    '5.18', '5.19', '6.0', '6.1', '6.2', '6.3', '6.4', '6.5',
                    '6.6', '6.7', '6.8', '6.9', '6.10', '6.11', '6.12',
-                   '6.13', '6.14', '6.15', '6.16', '6.17']
+                   '6.13', '6.14', '6.15', '6.16', '6.17', '6.18', '6.19', '7.0']
 
 def parse_args():
     parser = argparse.ArgumentParser('Script to populate the syscalls.csv kernel versions',
@@ -66,6 +67,10 @@ def parse_args():
     else:
         args.versions = args.versions.split(',')
 
+    # Convert to absolute paths
+    args.datapath = Path(args.datapath).resolve()
+    args.kernelpath = Path(args.kernelpath).resolve()
+
     return args
 
 def run(command, verbose=False, shell=False, timeout=None):
@@ -135,10 +140,10 @@ def main(args):
 
         src_path = os.path.join(args.datapath, 'data/tables')
         dest_path = os.path.join(os.getcwd(), 'tables-{}'.format(kver))
-        cp_cmd = 'cp -r {} {}'.format(src_path, dest_path)
+        cp_cmd = 'cp -r {}/. {}'.format(src_path, dest_path)
         ret, out, err = run(cp_cmd, shell=True)
         if ret != 0:
-            raise RuntimeError('Table copy failed: {}'.format(ret))
+            raise RuntimeError('Table copy failed: {}'.format(err))
 
 if __name__ == '__main__':
     args = parse_args()

src/arch-update-syscalls-csv.py

@@ -21,6 +21,7 @@
 # along with this library; if not, see <http://www.gnu.org/licenses>.
 #
 
+from pathlib import Path
 import subprocess
 import datetime
 import argparse
@@ -36,7 +37,7 @@
 ignore_syscall_list = [
     'arc_gettls', 'arc_settls', 'arc_usr_cmpxchg', 'bfin_spinlock',
     'cache_sync', 'clone2', 'cmpxchg_badaddr', 'dipc', 'dma_memcpy',
-    'exec_with_loader', 'execv', 'file_getattr', 'file_setattr',
+    'exec_with_loader', 'execv',
     'flush_cache', 'fp_udfiex_crtl', 'getdomainname', 'getdtablesize',
     'gethostname', 'getunwind', 'getxgid', 'getxpid', 'getxuid',
     'kern_features', 'llseek', 'madvise1', 'memory_ordering', 'metag_get_tls',
@@ -79,6 +80,10 @@ def parse_args():
     args = parser.parse_args()
     args.versions = args.versions.split(',')
 
+    # Convert to absolute paths
+    args.datapath = Path(args.datapath).resolve()
+    args.kernelpath = Path(args.kernelpath).resolve()
+
     return args
 
 def run(command, verbose=False, shell=False, timeout=None):
@@ -182,18 +187,25 @@ def parse_syscalls_csv(args):
 
 def insert_new_syscall(syscalls, syscall_name, column_cnt):
     inserted = False
+    new_row = list()
+
+    for i in range(0, column_cnt):
+        if (i % 2) == 0:
+            new_row.append('PNR')
+        else:
+            new_row.append('SCMP_KV_UNDEF')
 
     for syscall in syscalls:
         if syscall_name < syscall:
             idx = list(syscalls.keys()).index(syscall)
             syscalls_list = list(syscalls.items())
-            syscalls_list.insert(idx, (syscall_name, ['PNR'] * column_cnt))
+            syscalls_list.insert(idx, (syscall_name, new_row))
             syscalls = dict(syscalls_list)
             inserted = True
             break
 
     if not inserted:
-        syscalls[syscall_name] = ['PNR'] * column_cnt
+        syscalls[syscall_name] = new_row
 
     return syscalls
 
@@ -290,8 +302,15 @@ def update_syscalls_dict(args, columns, syscalls, kver):
                                   format(syscall_name, column, kver))
 
                         syscalls[syscall_name][col_idx] = str(syscall_num)
-                        maj = kver.split('.')[0]
-                        mnr = kver.split('.')[1]
+
+                        if kver.find('-rc') > 0:
+                            # Remove trailing release candidate tags
+                            tmp_kver = kver.split('-')[0]
+                        else:
+                            tmp_kver = kver
+
+                        maj = tmp_kver.split('.')[0]
+                        mnr = tmp_kver.split('.')[1]
                         syscalls[syscall_name][col_idx + 1] = \
                             'SCMP_KV_{}_{}'.format(maj, mnr)