Bugfix: MODE SELECT for format page is incorrect (issue #818)

[uweseimet]

Successfully tested with software that can change the sector size before formatting. Unit tests were added for the updated code.

After fixing the descriptor and offset handling rascsi behaves like a Fujitsu M2624S drive and a MODE SELECT/FORMAT sequence works as far as rascsi can support it.

[2022-10-08 13:07:25.373] [debug] ++++ CMD ++++ void ScsiController::Execute() Executing command $1A
[2022-10-08 13:07:25.373] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = ModePageDevice] Executing ModeSense6 ($1A)
[2022-10-08 13:07:30.382] [debug] ++++ CMD ++++ void ScsiController::Execute() Executing command $1A
[2022-10-08 13:07:30.382] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = ModePageDevice] Executing ModeSense6 ($1A)
[2022-10-08 13:07:30.384] [debug] ++++ CMD ++++ void ScsiController::Execute() Executing command $15
[2022-10-08 13:07:30.384] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = ModePageDevice] Executing ModeSelect6 ($15)
[2022-10-08 13:07:30.421] [debug] ++++ CMD ++++ void ScsiController::Execute() Executing command $04
[2022-10-08 13:07:30.421] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = Disk] Executing FormatUnit ($04)


2022-10-08 13:08:47.508] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = ModePageDevice] Executing ModeSense6 ($1A)
[2022-10-08 13:08:54.682] [debug] ++++ CMD ++++ void ScsiController::Execute() Executing command $1A
[2022-10-08 13:08:54.682] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = ModePageDevice] Executing ModeSense6 ($1A)
[2022-10-08 13:08:54.684] [debug] ++++ CMD ++++ void ScsiController::Execute() Executing command $15
[2022-10-08 13:08:54.684] [debug] bool Dispatcher<T>::Dispatch(T*, scsi_defs::scsi_command) [with T = ModePageDevice] Executing ModeSelect6 ($15)
[2022-10-08 13:08:54.684] [warning] In order to change the sector size use the -b option when launching rascsi
[2022-10-08 13:08:54.685] [debug] Error status: Sense Key $05, ASC $26

[akuker]

Approved @uweseimet . I'll let you do the merging. Thanks!

[rdmark]

This might not have been caused by the changes in this PR, but I'm getting this compiler error with gcc 10.2.1 / Bullseye / ARM (Pi Zero W).
Please let me know if you want an issue ticket with more context.

g++ -O3 -Wall -Werror -Wextra -DNDEBUG -Wno-psabi -std=c++17 -iquote . -D_FILE_OFFSET_BITS=64 -D_LARGEFILE64_SOURCE -MD -MP  -DFMT_HEADER_ONLY -DCONNECT_TYPE_FULLSPEC -c ./devices/device_factory.cpp -o obj/fullspec/device_factory.o
./devices/device_factory.cpp: In member function ‘std::__cxx11::list<std::__cxx11::basic_string<char> > DeviceFactory::GetNetworkInterfaces() const’:
./devices/device_factory.cpp:223:17: error: ‘char* strncpy(char*, const char*, size_t)’ specified bound 16 equals destination size [-Werror=stringop-truncation]
  223 |          strncpy(ifr.ifr_name, tmp->ifa_name, IFNAMSIZ); //NOSONAR Using strncpy is safe here
      |          ~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1plus: all warnings being treated as errors
make: *** [Makefile:156: obj/fullspec/device_factory.o] Error 1

[uweseimet]

@akuker Looks as if your suggested change (which I liked better than the previous code) is causing this issue. I will change the code back to using strcpy as part of this PR. I hope that's fine for you.

[uweseimet]

@rdmark I fixed the compilation issue (which was depending on the compiler used) and double-checked that it compiles again on bullseye.

[sonarcloud]

SonarCloud Quality Gate failed.   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

64.7% Coverage
0.0% Duplication

[uweseimet]

@akuker @rdmark I think I know now how to make the compiler happy while still using strncpy:

strncpy(ifr.ifr_name, tmp->ifa_name, IFNAMSIZ - 1)

Actually the compiler is right: If you copy exactly IFNAMSIZ bytes the terminating null byte for the string will be missing.
We can change the code accordingly, or we can keep it as it is. Both solutions are safe.