Modify get_time_info() to return seconds and nanoseconds

[visitorckw]

Modify get_time_info() to return seconds and nanoseconds to get higher time resolution. Additionally, remove the outdated TODO comments as it was solved in newlib v4.4.0 [1].

Link: https://sourceware.org/git/?p=newlib-cygwin.git;a=commit;h=5f15d7c5817b07a6b18cbab17342c95cb7b42be4 [1]

[jserv]

Thank @visitorckw for contributing!

[vacantron]

@visitorckw, I found the results of some tests like dhrystone and coremark being weird after applying the changes of timespec. Should we also update the binary to reflect the change of newlib?

[visitorckw]

@visitorckw, I found the results of some tests like drystone and coremark being weird after applying the changes of timespec. Should we also update the binary to reflect the change of newlib?

Absolutely. When I discovered the errors in newlib libgloss, @jserv asked me to correct the time-related system calls in rv32emu and also update all the binaries. However, I remember that when I applied my newlib patch and compiled the riscv gnu toolchain to generate the test binaries for rv32emu, I encountered an undefined reference to `nanosleep' error. Since I couldn't figure out the cause of this error at that time, I haven't made this update.

[visitorckw]

@visitorckw, I found the results of some tests like drystone and coremark being weird after applying the changes of timespec. Should we also update the binary to reflect the change of newlib?

Additionally, I believe the weird results you observed are due to the change in time units introduced by PR #360 , not this PR. This PR merely optimized the precision without changing the time units.

[vacantron]

Additionally, I believe the weird results you observed are due to the change in time units introduced by PR #360 , not this PR. This PR merely optimized the precision without changing the time units.

Yes, I have found that I replied the wrong one. I sorted the PRs with created time and not realized it. 🥲

[visitorckw]

Yes, I have found that I replied the wrong one. I sorted the PRs with created time and not realized it. 🥲

That's fine. Replying to either one is fine.
I just tried using the updated riscv gnu toolchain to compile nyancat.c and still encountered the following error:

nyancat.c: In function 'main':
nyancat.c:940:9: warning: implicit declaration of function 'nanosleep' [-Wimplicit-function-declaration]
  940 |         nanosleep(&ts, &ts);
      |         ^~~~~~~~~
in function `main':
nyancat.c:(.text+0x34c): undefined reference to `nanosleep'
collect2: error: ld returned 1 exit status

Do you have any idea about this error?

[visitorckw]

That's fine. Replying to either one is fine. I just tried using the updated riscv gnu toolchain to compile nyancat.c and still encountered the following error:

nyancat.c: In function 'main':
nyancat.c:940:9: warning: implicit declaration of function 'nanosleep' [-Wimplicit-function-declaration]
  940 |         nanosleep(&ts, &ts);
      |         ^~~~~~~~~
in function `main':
nyancat.c:(.text+0x34c): undefined reference to `nanosleep'
collect2: error: ld returned 1 exit status

Do you have any idea about this error?

I encountered this error back in November or December of last year. This issue has indeed been pending for quite a while without resolution. If we still can't identify the cause of this error, perhaps I could write a script to perform a bisection on the riscv gnu toolchain to try and find the cause.

[vacantron]

I encountered this error back in November or December of last year. This issue has indeed been pending for quite a while without resolution. If we still can't identify the cause of this error, perhaps I could write a script to perform a bisection on the riscv gnu toolchain to try and find the cause.

I add -lsemihost to the compilation option and the problem is solved. It seems they turned some libgloss's static linked libraries into dynamic ones.