From Chenguang Wang to ~mpu/qbe
I don't think so:
qbe $ ./qbe test/tls.ssa | tail -n17
.text
xvalcnt:
.cfi_startproc
pushq %rbp
movq %rsp, %rbp
movq %fs:0, %rax
leaq x@tpoff(%rax), %rax
movl (%rax, %rdi, 4), %eax
leave
ret
From Chenguang Wang to ~mpu/qbe
This reverts commit 4bc4c9584a13736c20855cdea2203d3bd0a259a3.
We can re-enable BTI now, since the stock `as` in OpenBSD 7.5 seems to
support it:
qbe $ ./qbe test/tls.ssa > /tmp/tls.s
qbe $ grep endbr64 /tmp/tls.s | uniq -c
6 endbr64
qbe $ as /tmp/tls.s -o /tmp/tls.o
qbe $ ls -al /tmp/tls.o
-rw-r--r-- 1 w wheel 2312 Apr 19 06:57 /tmp/tls.o
isel4.ssa and tls.ssa currently fail with SIGILL on my machine. This
commit fixes that.
[message trimmed]
From Chenguang Wang to ~sircmpwn/hare-dev
In testmod/testmod.ha, s_x/s_a was defined as a prototype [1]:
export let @symbol("s_x") s_a: int;
But tests/34-declarations.ha uses it [2]:
fn imported() void = {
testmod::s_a;
}
This currently works on QBE because QBE recognizes the `loadsw $s_x`
return value is never used, and eliminates the expression, but in
fact $s_x is never defined anywhere in the final linked binary
tests_34_declarations.
[message trimmed]
From Chenguang Wang to ~mpu/qbe
Fair points. Just for the record...
1. `-Wsizeof-pointer-memaccess` catches `sizeof` misuses on pointer
types, but it only works on gcc; the clang version doesn't detect this:
$ cat t.c
#include <stdio.h>
int main(void) {
char *p = 0;
char buf[42];
snprintf(p, sizeof p, "hi");
snprintf(buf, sizeof buf, "hi");
}
From Chenguang Wang to ~mpu/qbe
Hi, This is irrelevant to code size or runtime overhead, but I do wish this feature could be controlled behind a flag, because on OpenBSD the default binutils shipped with system is 2.17, which does not seem to support `endbr64`. I had to 1) run `pkg_add binutils` and 2) configure my copy of cproc using `--with-as=gas` to make it work as expected. It would be great if QBE could remove non-stock binutils as a hard dependency on OpenBSD. On Sat, Dec 30, 2023, at 7:19 AM, Quentin Carbonneaux wrote: > Hi all, >
From Chenguang Wang to ~mpu/qbe
See previous discussion here[1]. It's unclear to me how to preview HTML page generated from the .txt file, though. [1]: https://lists.sr.ht/~mpu/qbe/%3C87cz1jq26k.fsf%40greenfork.me%3E --- doc/il.txt | 30 ++++++++++++++++++++++++++++-- 1 file changed, 28 insertions(+), 2 deletions(-) diff --git a/doc/il.txt b/doc/il.txt index cc3e021..b036273 100644 --- a/doc/il.txt +++ b/doc/il.txt @@ -15,7 +15,7 @@ 2. <@ Types > [message trimmed]
From Chenguang Wang to ~mpu/qbe
When compiling on OpenBSD, the compiler produces warning messages about the usage of these two functions, and suggests replacing them with snprintf() and strlcpy(). strlcpy() on Linux requires including bsd/string.h and linking libbsd. I found it easier to just use equivalent snprintf() calls instead. --- amd64/emit.c | 2 +- amd64/isel.c | 2 +- arm64/emit.c | 10 +++++----- arm64/isel.c | 2 +- minic/minic.y | 21 +++++++++++---------- minic/yacc.c | 24 ++++++++++++------------ parse.c | 8 ++++---- [message trimmed]
From Chenguang Wang to ~mpu/qbe
> VAL could also be $IDENT, e.g. in function calls:
My apologies -- apparently $IDENT is CONST already.
From Chenguang Wang to ~mpu/qbe
Just found this thread while writing my own QBE parser. > > 2. I haven't found the definition of VAL, is it true that > > VAL := DYNCONST | %IDENT > > > Yes that's a valid definition I believe. VAL is what the > parseref() function parses. > > See https://c9x.me/git/qbe.git/tree/parse.c#n402 VAL could also be $IDENT, e.g. in function calls: // CALL := [%IDENT '=' ABITY] 'call' VAL '(' (ARG), ')'