[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20190109131131.GB25192@kernel.org>
Date: Wed, 9 Jan 2019 10:11:31 -0300
From: Arnaldo Carvalho de Melo <acme@...nel.org>
To: "Michael S. Tsirkin" <mst@...hat.com>
Cc: Ingo Molnar <mingo@...nel.org>,
Clark Williams <williams@...hat.com>,
linux-kernel@...r.kernel.org, linux-perf-users@...r.kernel.org,
Arnaldo Carvalho de Melo <acme@...hat.com>,
Adrian Hunter <adrian.hunter@...el.com>,
Jiri Olsa <jolsa@...nel.org>,
Luis Cláudio Gonçalves <lclaudio@...hat.com>,
Namhyung Kim <namhyung@...nel.org>,
Paolo Bonzini <pbonzini@...hat.com>,
Wang Nan <wangnan0@...wei.com>
Subject: Re: [PATCH 16/16] tools include uapi: Sync linux/vhost.h with the
kernel sources
Em Tue, Jan 08, 2019 at 11:10:10PM -0500, Michael S. Tsirkin escreveu:
> On Tue, Jan 08, 2019 at 04:59:10PM -0300, Arnaldo Carvalho de Melo wrote:
> > +++ b/tools/include/uapi/linux/vhost.h
> > @@ -11,94 +11,9 @@
> > * device configuration.
> > */
> > +#include <linux/vhost_types.h>
> Don't you then also need to import vhost_types.h?
If we were using this file as an include in a C source file, yes, but
we're just using it to generate a string table using regular
expressions:
[acme@...co perf]$ tools/perf/trace/beauty/vhost_virtio_ioctl.sh
static const char *vhost_virtio_ioctl_cmds[] = {
[0x00] = "SET_FEATURES",
[0x01] = "SET_OWNER",
[0x02] = "RESET_OWNER",
[0x03] = "SET_MEM_TABLE",
[0x04] = "SET_LOG_BASE",
[0x07] = "SET_LOG_FD",
[0x10] = "SET_VRING_NUM",
[0x11] = "SET_VRING_ADDR",
[0x12] = "SET_VRING_BASE",
[0x13] = "SET_VRING_ENDIAN",
[0x14] = "GET_VRING_ENDIAN",
[0x20] = "SET_VRING_KICK",
[0x21] = "SET_VRING_CALL",
[0x22] = "SET_VRING_ERR",
[0x23] = "SET_VRING_BUSYLOOP_TIMEOUT",
[0x24] = "GET_VRING_BUSYLOOP_TIMEOUT",
[0x25] = "SET_BACKEND_FEATURES",
[0x30] = "NET_SET_BACKEND",
[0x40] = "SCSI_SET_ENDPOINT",
[0x41] = "SCSI_CLEAR_ENDPOINT",
[0x42] = "SCSI_GET_ABI_VERSION",
[0x43] = "SCSI_SET_EVENTS_MISSED",
[0x44] = "SCSI_GET_EVENTS_MISSED",
[0x60] = "VSOCK_SET_GUEST_CID",
[0x61] = "VSOCK_SET_RUNNING",
};
static const char *vhost_virtio_ioctl_read_cmds[] = {
[0x00] = "GET_FEATURES",
[0x12] = "GET_VRING_BASE",
[0x26] = "GET_BACKEND_FEATURES",
};
[acme@...co perf]$
[acme@...co perf]$ cat tools/perf/trace/beauty/vhost_virtio_ioctl.sh
#!/bin/sh
# SPDX-License-Identifier: LGPL-2.1
[ $# -eq 1 ] && header_dir=$1 || header_dir=tools/include/uapi/linux/
printf "static const char *vhost_virtio_ioctl_cmds[] = {\n"
regex='^#[[:space:]]*define[[:space:]]+VHOST_(\w+)[[:space:]]+_IOW?\([[:space:]]*VHOST_VIRTIO[[:space:]]*,[[:space:]]*(0x[[:xdigit:]]+).*'
egrep $regex ${header_dir}/vhost.h | \
sed -r "s/$regex/\2 \1/g" | \
sort | xargs printf "\t[%s] = \"%s\",\n"
printf "};\n"
printf "static const char *vhost_virtio_ioctl_read_cmds[] = {\n"
regex='^#[[:space:]]*define[[:space:]]+VHOST_(\w+)[[:space:]]+_IOW?R\([[:space:]]*VHOST_VIRTIO[[:space:]]*,[[:space:]]*(0x[[:xdigit:]]+).*'
egrep $regex ${header_dir}/vhost.h | \
sed -r "s/$regex/\2 \1/g" | \
sort | xargs printf "\t[%s] = \"%s\",\n"
printf "};\n"
[acme@...co perf]$
This ends up being used in tools/perf/trace/beauty/ioctl.c, after that
trace/beauty/generated/ioctl/vhost_virtio_ioctl_array.c file (with the
vhost_virtio_ioctl_read_cmds and vhost_virtio_ioctl_cmds arrays) gets
generated by the tools/perf/Makefile.perf build process:
static size_t ioctl__scnprintf_vhost_virtio_cmd(int nr, int dir, char *bf, size_t size)
{
#include "trace/beauty/generated/ioctl/vhost_virtio_ioctl_array.c"
static DEFINE_STRARRAY(vhost_virtio_ioctl_cmds, "");
static DEFINE_STRARRAY(vhost_virtio_ioctl_read_cmds, "");
struct strarray *s = (dir & _IOC_READ) ? &strarray__vhost_virtio_ioctl_read_cmds : &strarray__vhost_virtio_ioctl_cmds;
if (nr < s->nr_entries && s->entries[nr] != NULL)
return scnprintf(bf, size, "VHOST_%s", s->entries[nr]);
return scnprintf(bf, size, "(%#x, %#x, %#x)", 0xAF, nr, dir);
}
So, if at some point a tool needs to really #include that file, yeah,
then we will need to import vhost_types.h too.
- Arnaldo
Powered by blists - more mailing lists