Re: [PATCH] perf arm64: Fix mksyscalltbl, don't lose syscalls due to sort -nu

From: Leo Yan
Date: Fri Nov 25 2022 - 07:54:13 EST


Hi Vincent,

[ + Arnd ]

On Fri, Nov 25, 2022 at 12:53:10PM +0100, Vincent Whitchurch wrote:
> On Tue, Dec 29, 2020 at 11:09:33AM +0800, Leo Yan wrote:
> > On Mon, Dec 28, 2020 at 03:39:41AM +0100, Hans-Peter Nilsson wrote:
> > > When using "sort -nu", arm64 syscalls were lost. That is, the
> > > io_setup syscall (number 0) and all but one (typically
> > > ftruncate; 64) of the syscalls that are defined symbolically
> > > (like "#define __NR_ftruncate __NR3264_ftruncate") at the point
> > > where "sort" is applied.
> > >
> > > This creation-of-syscalls.c-scheme is, judging from comments,
> > > copy-pasted from powerpc, and worked there because at the time,
> > > its tools/arch/powerpc/include/uapi/asm/unistd.h had *literals*,
> > > like "#define __NR_ftruncate 93".
> > >
> > > With sort being numeric and the non-numeric key effectively
> > > evaluating to 0, the sort option "-u" means these "duplicates"
> > > are removed. There's no need to remove syscall lines with
> > > duplicate numbers for arm64 because there are none, so let's fix
> > > that by just losing the "-u". Having the table numerically
> > > sorted on syscall-number for the rest of the syscalls looks
> > > nice, so keep the "-n".
> > >
> > > Signed-off-by: Hans-Peter Nilsson <hp@xxxxxxxx>
> >
> > Very good catching! I tested this patch with the commands:
> >
> > $ cd $LINUX_KERN
> > $ tools/perf/arch/arm64/entry/syscalls/mksyscalltbl \
> > $ARM64_TOOLCHAIN_PATH/aarch64-linux-gnu-gcc \
> > gcc tools tools/include/uapi/asm-generic/unistd.h
> >
> > It gives out complete syscall tables:
> >
> > $ diff /tmp/mksyscall_before.txt /tmp/mksyscall_after.txt
> > 1a2,4
> > > [223] = "fadvise64",
> > > [25] = "fcntl",
> > > [44] = "fstatfs",
> > 2a6,11
> > > [0] = "io_setup",
> > > [62] = "lseek",
> > > [222] = "mmap",
> > > [71] = "sendfile",
> > > [43] = "statfs",
> > > [45] = "truncate",
> >
> > Rather than dropping option "-u" for sort command, I googled and read
> > the manual of "sort", but cannot find other better method. So this
> > patch looks good for me:
> >
> > Reviewed-by: Leo Yan <leo.yan@xxxxxxxxxx>
> > Tested-by: Leo Yan <leo.yan@xxxxxxxxxx>
>
> It looks like this patch was never applied? AFAICS it is still needed
> on current HEAD and it still applies cleanly.

Thanks a lot for bringing up this.

Before there have a discussion [1] for refactoring Arm64 system call
table but it didn't really happen. I think it's the right thing to merge
this patch, @Arnaldo, could you pick up this patch?

Thanks,
Leo

[1] https://lore.kernel.org/lkml/CAK8P3a1G2YHh2FRd=vBY4xxsJ1wZFh2pB1wyw82QmQOaiV4+CA@xxxxxxxxxxxxxx/