syscall_emul: fix bugs for mmap2 system call and x86-32 syscalls

Information
Submitter:	Brandon Potter
Repository:	gem5
Branch:	default
Bugs:
Depends On:
Reviewers
Groups:	Default
People:

Description

Changeset 11343:aba7be5b02df
---------------------------
syscall_emul: fix bugs for mmap2 system call and x86-32 syscalls

Testing Done

Issue Summary

Description	From	Last Updated	Status
Is this needed for this patch?	Jason Lowe-Power	Feb. 11, 2016, 10:18 a.m.	Dropped
Did you mean to remove this DPRINTF?	Jason Lowe-Power	Feb. 11, 2016, 10:20 a.m.	Dropped
From what I can tell from manpages, this is always 4096 bytes. Are we assuming this is the page size ...	Jason Lowe-Power	Feb. 11, 2016, 3:21 p.m.	Dropped

src/arch/x86/process.cc (Diff revision 1)

Is this needed for this patch?

Show all issues

Brandon Potter Feb. 11, 2016, 10:43 a.m. (Feb. 11, 2016, 10:43 a.m.)

It's needed because mmap requires 6 arguments and only five registers were coded into the x86-32 model. (Linux requires up to 6 registers on x86 and EBP is the 6th: http://stackoverflow.com/questions/2535989/what-are-the-calling-conventions-for-unix-linux-system-calls-on-x86-64.)

Jason Lowe-Power Feb. 11, 2016, 3:21 p.m. (Feb. 11, 2016, 3:21 p.m.)
```
I see, thanks.
```

src/sim/syscall_emul.hh (Diff revision 1)

Did you mean to remove this DPRINTF?

Show all issues

Brandon Potter Feb. 11, 2016, 10:43 a.m. (Feb. 11, 2016, 10:43 a.m.)

Yes I did, it's redundant because the information is already printed out with the trace from the doSyscall function in syscall_emul.cc. From my perspective, this print doesn't add anything.

Jason Lowe-Power Feb. 11, 2016, 3:21 p.m. (Feb. 11, 2016, 3:21 p.m.)
```
Makes sense.
```

src/sim/syscall_emul.hh (Diff revision 1)

From what I can tell from manpages, this is always 4096 bytes. Are we assuming this is the page size of all of our ISAs, or are the manpages I'm looking at just x86 specific?

Show all issues

Brandon Potter Feb. 11, 2016, 10:43 a.m. (Feb. 11, 2016, 10:43 a.m.)

My manpage says the following: "offset must be a multiple of the page size as returned by sysconf(_SC_PAGE_SIZE)".

As you pointed out, this is 4096 for x86. Do you anticipate a problem with using TheISA::PageBytes? I'd rather not hardcode in 4096 if that's what you're suggesting.

Jason Lowe-Power Feb. 11, 2016, 3:21 p.m. (Feb. 11, 2016, 3:21 p.m.)

I think it's fine to keep it as PageBytes. I just wanted to make surethat somewhere the manpage said page size ;). Mine says "the final argument specifies the offset into the file in 4096-byte units". Which is similar to http://linux.die.net/man/2/mmap2

You have a pending review.

Review Board 2.0.15

This change has been marked as submitted.

Screenshots

Files

Issue Summary

Status: Closed (submitted)