feat: Added cheribsd preset by mereacre · Pull Request #358 · nqminds/edgesec

mereacre · 2022-12-06T14:38:29Z

Adds a new cheribsd preset with a different set of C flags specific for Cheri architecture in purecap mode.

codecov · 2022-12-06T14:48:59Z

Codecov Report

Merging #358 (de84d93) into main (4d5daaa) will increase coverage by 0.19%.
The diff coverage is 33.33%.

@@            Coverage Diff             @@
##             main     #358      +/-   ##
==========================================
+ Coverage   51.79%   51.99%   +0.19%     
==========================================
  Files         139      138       -1     
  Lines       19315    19239      -76     
==========================================
- Hits        10005    10004       -1     
+ Misses       9310     9235      -75

Impacted Files	Coverage Δ
src/runctl.c	`53.49% <33.33%> (-0.44%)`	⬇️
tests/test_edgesec.c	`94.53% <0.00%> (-0.79%)`	⬇️
src/sqlhook.c

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

aloisklink

This PR is referencing a CMakeModules/CMakeToolchains/cheri.cmake file, but it's not in this PR.

CMakePresets.json

Co-authored-by: Alois Klink <alois@nquiringminds.com>

aloisklink · 2022-12-06T15:29:56Z

CMakeModules/CMakeToolchains/cheri.cmake

+set(CMAKE_C_COMPILER clang)
+set(CMAKE_CXX_COMPILER clang++)
+
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -target aarch64-unknown-freebsd -march=morello+c64 -mabi=purecap -Xclang -morello-vararg=new")


We many want to adapt the official recommended cheri toolchain file from https://github.com/CTSRD-CHERI/cheribuild/blob/a1bdfa225b575606f7303435c4c75ea9942d697f/pycheribuild/files/CrossToolchain.cmake.in

aloisklink

I'll trust you on that the toolchain file works properly :)

I might need to wait until Friday until I get to my home computer to test this.

aloisklink · 2022-12-06T18:09:22Z

conftest.c

+/* confdefs.h */
+#define PACKAGE_NAME "util-linux"
+#define PACKAGE_TARNAME "util-linux"
+#define PACKAGE_VERSION "UNKNOWN"
+#define PACKAGE_STRING "util-linux UNKNOWN"
+#define PACKAGE_BUGREPORT "kzak@redhat.com"
+#define PACKAGE_URL "http://www.kernel.org/pub/linux/utils/util-linux/"
+/* end confdefs.h.  */
+
+int
+main (void)
+{
+
+  ;
+  return 0;
+}


I'm guessing this isn't used, so we can delete it.

aloisklink

Looks good! Congratulations on getting it working.

aloisklink · 2022-12-09T10:49:50Z

Hmmmm, a bit weird. It looks like the Debian build is failing to the following error:

[ 23%] Linking C executable test_wrap_log_error
cd /build/edgesec-0.1.0-alpha.0/obj-x86_64-linux-gnu/tests/utils && /usr/bin/cmake -E cmake_link_script CMakeFiles/test_wrap_log_error.dir/link.txt --verbose=1
/usr/bin/cc -g -O2 -ffile-prefix-map=/build/edgesec-0.1.0-alpha.0=. -flto=auto -ffat-lto-objects -flto=auto -ffat-lto-objects -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -Wunused-variable -Wall -Wextra -O3 -DNDEBUG -Wl,-Bsymbolic-functions -flto=auto -ffat-lto-objects -flto=auto -Wl,-z,relro -Wl,--wrap=log_levels CMakeFiles/test_wrap_log_error.dir/test_wrap_log_error.c.o CMakeFiles/wrap_log_error.dir/wrap_log_error.c.o -o test_wrap_log_error  ../../src/utils/liblog.a /usr/lib/x86_64-linux-gnu/libcmocka.so ../../src/utils/liblog.a /usr/lib/x86_64-linux-gnu/libcmocka.so 
/usr/bin/ld: /tmp/ccM8V57b.ltrans0.ltrans.o: in function `__wrap_log_levels':
./obj-x86_64-linux-gnu/tests/utils/./tests/utils/wrap_log_error.c:27: undefined reference to `log_levels'

I'm not 100% sure why, since I'm pretty sure the only thing you're changing is this line

edgesec/CMakeLists.txt

Line 161 in 252d883

set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wunused-variable -Wall -Wextra")

Edit: I've made a new branch called build/fix-CMAKE_C_FLAGS (commit 4b663a0) that only changes the CMAKE_C_FLAGS! Hopefully that should fix this issue (at least, once I get it working). Edit2: PR here #364

CMakeModules/CMakeToolchains/cheri-cross.cmake

Co-authored-by: Alois Klink <alois@nquiringminds.com>

aloisklink · 2023-02-03T11:21:41Z

By the way, @mereacre, this is the test errors that we're getting on the freebsd preset:

me@me:~/edgesec$ ctest --preset freebsd --output-on-failure --exclude-regex "$exclude_regex"
...
 2/40 Test  #2: test_runctl ......................***Failed    0.04 sec
[==========] tests: Running 2 test(s).
[ RUN      ] test_init_context
2022-12-13 13:50:16.697  �[36mDEBUG�[0m�[0m �[90mrunctl.c:257:�[0m Getting system commands paths
2022-12-13 13:50:16.697  �[36mDEBUG�[0m�[0m �[90mrunctl.c:321:�[0m Opening the macconn db=:memory:
2022-12-13 13:50:16.718  �[36mDEBUG�[0m�[0m �[90mrunctl.c:328:�[0m Creating subnet to interface mapper...
2022-12-13 13:50:16.718  �[36mDEBUG�[0m�[0m �[90mrunctl.c:339:�[0m Creating VLAN ID to interface mapper...
[       OK ] test_init_context
[ RUN      ] test_run_engine
2022-12-13 13:50:16.718  �[36mDEBUG�[0m�[0m �[90mrunctl.c:257:�[0m Getting system commands paths
2022-12-13 13:50:16.718  �[36mDEBUG�[0m�[0m �[90mrunctl.c:321:�[0m Opening the macconn db=:memory:
2022-12-13 13:50:16.719  �[36mDEBUG�[0m�[0m �[90mrunctl.c:328:�[0m Creating subnet to interface mapper...
2022-12-13 13:50:16.719  �[36mDEBUG�[0m�[0m �[90mrunctl.c:339:�[0m Creating VLAN ID to interface mapper...
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:430:�[0m AP name: 
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:431:�[0m AP interface: 
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:454:�[0m Setting the ip forward os system flag...
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:461:�[0m Adding default mac mappers...
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:468:�[0m Creating subnet interfaces...
2022-12-13 13:50:16.719  �[36mDEBUG�[0m�[0m �[90miface.c:284:�[0m Commiting interface changes
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:479:�[0m Creating supervisor on  with port 0
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:488:�[0m Looking for VLAN capable wifi interface...
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:520:�[0m Creating the radius server on port 0 with client ip 
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:530:�[0m Running the dhcp service...
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:541:�[0m Running the mdns forwarder service thread...
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:551:�[0m ++++++++++++++++++
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:552:�[0m Running event loop
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:553:�[0m ++++++++++++++++++
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:556:�[0m Exit event loop
2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90meloop.c:655:�[0m eloop_get_sock_table: eloop data ptr was NULL
[  ERROR   ] --- Test failed with exception: Bus error(10)
[  FAILED  ] test_run_engine
[==========] tests: 2 test(s) run.
[  PASSED  ] 1 test(s).
[  FAILED  ] tests: 1 test(s), listed below:
[  FAILED  ] test_run_engine

 1 FAILED TEST(S)

I'm guessing it is to do with how you changed ip to ifconfig in https://github.com/nqminds/edgesec/blob/de84d93d38bfd843770ac36e131090d1c1845222/src/runctl.c, but I don't know enough about 🚌 to know why there is a 🚌 error.

Maybe we have an uninitialized pointer somewhere, or there are alignment issues?

mereacre · 2023-02-03T11:26:45Z

 Bus Error (also known as SIGBUS and is usually signal 10) occur when a process is trying to access memory that the CPU cannot physically address...

I believe that after

2022-12-13 13:50:16.719  �[32mINFO �[0m�[0m �[90mrunctl.c:556:�[0m Exit event loop

runctl is trying to free up memory and accessed something that was already freed and therefore it gives the Bus(10) error. We need to replicate this error on other presets.

aloisklink · 2023-02-03T11:35:10Z

runctl is trying to free up memory and accessed something that was already freed and therefore it gives the Bus(10) error. We need to replicate this error on other presets.

Hmmm, I think we don't actually ever test the "USE_GENERIC_IP_SERVICE": true syntax except for the freebsd preset.

The recap preset does use USE_GENERIC_IP_SERVICE:

edgesec/CMakePresets.json

Lines 74 to 83 in e31c8c8

    
           "name": "recap", 
        
           "inherits": "default", 
        
           "displayName": "recap tool", 
        
           "description": "recap tool", 
        
           "cacheVariables": { 
        
             "BUILD_MNL_LIB": false, 
        
             "BUILD_NETLINK_LIB": false, 
        
             "USE_NETLINK_SERVICE": false, 
        
             "USE_UCI_SERVICE": false, 
        
             "USE_GENERIC_IP_SERVICE": true,

However, there is a recap build preset, but no recap test preset.

mereacre · 2023-02-03T12:04:34Z

The USE_GENERIC_IP_SERVICE uses the ip command (by executing the process) instead of libnl. So, that shouldn't work on freebsd because it uses ifconfig.

aloisklink · 2023-02-03T12:09:54Z

The tests were passing previously, though. Maybe edgesec had better error handling for when ip wasn't found, but doesn't know what to do with ifconfig?

My gut feeling is to undo the src/runctl.c changes so that we can merge this PR, and maybe make them in a new PR called "Use ifconfig for ip service", but it's up to you.

Edit: I usually use https://github.com/CTSRD-CHERI/cheribuild to do cross-compiling, so I'm probably not going to use this cheribsd preset.

mereacre · 2023-02-03T12:26:23Z

Which changes do you want to undo?

The USE_GENERIC_IP_SERVICE was added a year ago as I remember.

aloisklink · 2023-02-03T12:33:41Z

Which changes do you want to undo?

These changes:

https://github.com/nqminds/edgesec/pull/358/files#diff-2952f7788619c2fe81d6536e6b254d97a9e46b87e2f7d54a4b7e15e5624aad6eR245-R270

Essentially, we should keep src/runctl.c how it is on the main branch, and not change it in this PR. We can then update src/runctl.c to work with ifconfig in a separate PR.

mereacre

Ok

feat: Added cheribsd preset

7352247

mereacre requested a review from aloisklink December 6, 2022 14:38

aloisklink reviewed Dec 6, 2022

View reviewed changes

CMakePresets.json Show resolved Hide resolved

mereacre and others added 2 commits December 6, 2022 14:59

chore: cleanup

c17469d

Update CMakePresets.json

64a9bb7

Co-authored-by: Alois Klink <alois@nquiringminds.com>

aloisklink reviewed Dec 6, 2022

View reviewed changes

feat: Added a new target cherbsd cross compile

0233264

aloisklink approved these changes Dec 6, 2022

View reviewed changes

feat: Added CFLAGS to libs

88ca70e

mereacre requested a review from aloisklink December 6, 2022 18:09

aloisklink reviewed Dec 6, 2022

View reviewed changes

aloisklink approved these changes Dec 6, 2022

View reviewed changes

mereacre added 2 commits December 9, 2022 10:32

chore: cleanup

da18f74

fix: removed the test preset for cheribsd-cc

252d883

mereacre and others added 6 commits December 9, 2022 11:11

Merge branch 'main' into cheri-pure

bdfec7f

featr: Added search for ifconfig

cbeb718

feat: Added option to build sqlhook

12460ed

feat: Added preset options

8f65db9

Merge branch 'main' into cheri-pure

b778871

Merge branch 'main' into cheri-pure

28ed1b1

mereacre added this to the CheriBSD Support milestone Dec 12, 2022

Merge branch 'main' into cheri-pure

17451c4

aloisklink reviewed Dec 12, 2022

View reviewed changes

CMakeModules/CMakeToolchains/cheri-cross.cmake Outdated Show resolved Hide resolved

mereacre and others added 5 commits December 12, 2022 13:23

Update CMakeModules/CMakeToolchains/cheri-cross.cmake

19b0ab9

Co-authored-by: Alois Klink <alois@nquiringminds.com>

Merge branch 'main' into cheri-pure

7bb52cf

chore: pre-commit

4854e60

Merge branch 'main' into cheri-pure

45fe2d7

Merge branch 'main' into cheri-pure

299e316

Merge branch 'main' into cheri-pure

de84d93

mereacre commented Feb 3, 2023

View reviewed changes

aloisklink mentioned this pull request Jun 2, 2023

Planned software infrastructure nqminds/trustnetz#4

Open

Conversation

mereacre commented Dec 6, 2022

Uh oh!

codecov bot commented Dec 6, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

aloisklink left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

aloisklink Dec 6, 2022

Choose a reason for hiding this comment

Uh oh!

aloisklink left a comment

Choose a reason for hiding this comment

Uh oh!

aloisklink Dec 6, 2022

Choose a reason for hiding this comment

Uh oh!

aloisklink left a comment

Choose a reason for hiding this comment

Uh oh!

aloisklink commented Dec 9, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

aloisklink commented Feb 3, 2023

Uh oh!

mereacre commented Feb 3, 2023

Uh oh!

aloisklink commented Feb 3, 2023

Uh oh!

mereacre commented Feb 3, 2023

Uh oh!

aloisklink commented Feb 3, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mereacre commented Feb 3, 2023

Uh oh!

aloisklink commented Feb 3, 2023

Uh oh!

mereacre left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

codecov bot commented Dec 6, 2022 •

edited

Loading

aloisklink commented Dec 9, 2022 •

edited

Loading

aloisklink commented Feb 3, 2023 •

edited

Loading