~mpu

http://c9x.me

~mpu/qbe

Last active 29 minutes ago
View more

Recent activity

Re: Crash in mem.c: coalesce: Assertion `u->type == UIns' failed. a day ago

From Quentin Carbonneaux to ~mpu/qbe

On Wed, Feb 21, 2024, at 08:18, Jummit wrote:
> If this is something I should report in the Hare project instead,
> please let me know.

This is precisely the right list. Thanks for reporting, I'll
look into it.

QBE 1.2 7 days ago

From Quentin Carbonneaux to ~mpu/qbe

QBE 1.2 is now released and available at:
    http://c9x.me/compile/releases.html

QBE 1.2 comes with important correctness fixes as well as
improvements to the documentation. Notable changes are:
  * correctness fixes to the stack-slot coalescing pass;
  * new experimental dbgfile and dbgloc directives.

Happy hacking!

Re: [PATCH] Print total number of tests run to get a better view of how much is broken. 25 days ago

From Quentin Carbonneaux to ~mpu/qbe

Eh, why not? I merged a slightly modified version in master.

Cheers.

Re: qbe fails to eliminate some stack slots a month ago

From Quentin Carbonneaux to ~mpu/qbe

Messaged you with some initial instructions.

Cheers.

Re: [PATCH] Fix sprintf() and strcpy() warnings on OpenBSD. a month ago

From Quentin Carbonneaux to ~mpu/qbe

On Sat, Dec 30, 2023, at 18:48, Chenguang Wang wrote:
> When compiling on OpenBSD, the compiler produces warning messages about the
> usage of these two functions, and suggests replacing them with snprintf() and
> strlcpy().

I have mixed feelings about that. Additionally all of the
`sizeof xxx` proposed will break silently if we move to
pointer types instead of array types.

I manually reviewed:

  - https://c9x.me/git/qbe.git/tree/amd64/emit.c#n194
    fine, the %d expands to at most 2 chars, the buf is 6 chars
    and fits "xmmNN". It's easy for the compiler to see that

Re: [PATCH] dbgloc: add column argument a month ago

From Quentin Carbonneaux to ~mpu/qbe

On Tue, Jan 2, 2024, at 11:20, Drew DeVault wrote:
> Even if we were to table that option, I'm not for it even without
> compatibility concerns. Some compilers may not bother collecting
> columns, and GNU as's .loc macro also makes the column optional. So it
> makes sense for it to be an optional parameter.

Okay, I just pushed a simplified version.
Let me know how that works for you.

Re: [PATCH] Add missing documentation for union types and VAL. a month ago

From Quentin Carbonneaux to ~mpu/qbe

Super work, thanks!

It's now merged in master.

Re: [PATCH] dbgloc: add column argument a month ago

From Quentin Carbonneaux to ~mpu/qbe

On Tue, Jan 2, 2024, at 11:09, Drew DeVault wrote:
> This is implemented in a backwards-compatible manner.

dbgloc is still experimental and undocumented,
what would you think of breaking backwards
compatibility?

Control-flow integrity overhead a month ago

From Quentin Carbonneaux to ~mpu/qbe

Hi all,

I merged 5af33410 in main today but I'm having second thoughts.
I was expecting the assembler can be put in a mode where the
control-flow integrity instructions added to each function
prelude can be eliminated, but it does not look like it is the
case. Linkers could be smart about it but I don't expect they
are today.

Please report any observed code-size / runtime overhead you go
by with qbe master. We can gate cfi behind a linkage flag it
proves to be a problem.

Cheers.

Re: [PATCH] Fix IBT/BTI by instrumenting function calls a month ago

From Quentin Carbonneaux to ~mpu/qbe

Thanks, it's applied and pushed to master.