On Wed, Apr 01, 2020 at 10:13:29PM +0800, Jason Wang wrote:

quoted

On 2020/4/1 下午9:02, Christian Borntraeger wrote:

quoted

On 01.04.20 14:56, Christian Borntraeger wrote:

quoted

On 01.04.20 14:50, Jason Wang wrote:

quoted

On 2020/4/1 下午7:21, Christian Borntraeger wrote:

quoted

On 26.03.20 15:01, Jason Wang wrote:

quoted

Currently, CONFIG_VHOST depends on CONFIG_VIRTUALIZATION. But vhost is
not necessarily for VM since it's a generic userspace and kernel
communication protocol. Such dependency may prevent archs without
virtualization support from using vhost.

To solve this, a dedicated vhost menu is created under drivers so
CONIFG_VHOST can be decoupled out of CONFIG_VIRTUALIZATION.

FWIW, this now results in vhost not being build with defconfig kernels (in todays
linux-next).

Hi Christian:

Did you meet it even with this commit https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=a4be40cbcedba9b5b714f3c95182e8a45176e42d?

I simply used linux-next. The defconfig does NOT contain CONFIG_VHOST and therefore CONFIG_VHOST_NET and friends
can not be selected.

$ git checkout next-20200401
$ make defconfig
    HOSTCC  scripts/basic/fixdep
    HOSTCC  scripts/kconfig/conf.o
    HOSTCC  scripts/kconfig/confdata.o
    HOSTCC  scripts/kconfig/expr.o
    LEX     scripts/kconfig/lexer.lex.c
    YACC    scripts/kconfig/parser.tab.[ch]
    HOSTCC  scripts/kconfig/lexer.lex.o
    HOSTCC  scripts/kconfig/parser.tab.o
    HOSTCC  scripts/kconfig/preprocess.o
    HOSTCC  scripts/kconfig/symbol.o
    HOSTCC  scripts/kconfig/util.o
    HOSTLD  scripts/kconfig/conf
*** Default configuration is based on 'x86_64_defconfig'
#
# configuration written to .config
#

$ grep VHOST .config
# CONFIG_VHOST is not set

quoted

If yes, what's your build config looks like?

Thanks

This was x86. Not sure if that did work before.
On s390 this is definitely a regression as the defconfig files
for s390 do select VHOST_NET

grep VHOST arch/s390/configs/*
arch/s390/configs/debug_defconfig:CONFIG_VHOST_NET=m
arch/s390/configs/debug_defconfig:CONFIG_VHOST_VSOCK=m
arch/s390/configs/defconfig:CONFIG_VHOST_NET=m
arch/s390/configs/defconfig:CONFIG_VHOST_VSOCK=m

and this worked with 5.6, but does not work with next. Just adding
CONFIG_VHOST=m to the defconfig solves the issue, something like

Right, I think we probably need

1) add CONFIG_VHOST=m to all defconfigs that enables
CONFIG_VHOST_NET/VSOCK/SCSI.

or

2) don't use menuconfig for CONFIG_VHOST, let NET/SCSI/VDPA just select it.

Thanks

OK I tried this:

diff --git a/drivers/vhost/Kconfig b/drivers/vhost/Kconfig
index 2523a1d4290a..a314b900d479 100644
--- a/drivers/vhost/Kconfig
+++ b/drivers/vhost/Kconfig

@@ -19,11 +19,10 @@ menuconfig VHOST
  	  This option is selected by any driver which needs to access
  	  the core of vhost.
  
-if VHOST
-
  config VHOST_NET
  	tristate "Host kernel accelerator for virtio net"
  	depends on NET && EVENTFD && (TUN || !TUN) && (TAP || !TAP)
+	select VHOST
  	---help---
  	  This kernel module can be loaded in host kernel to accelerate
  	  guest networking with virtio_net. Not to be confused with virtio_net

@@ -35,6 +34,7 @@ config VHOST_NET
  config VHOST_SCSI
  	tristate "VHOST_SCSI TCM fabric driver"
  	depends on TARGET_CORE && EVENTFD
+	select VHOST
  	default n
  	---help---
  	Say M here to enable the vhost_scsi TCM fabric module

@@ -44,6 +44,7 @@ config VHOST_VSOCK
  	tristate "vhost virtio-vsock driver"
  	depends on VSOCKETS && EVENTFD
  	select VIRTIO_VSOCKETS_COMMON
+	select VHOST
  	default n
  	---help---
  	This kernel module can be loaded in the host kernel to provide AF_VSOCK

@@ -57,6 +58,7 @@ config VHOST_VDPA
  	tristate "Vhost driver for vDPA-based backend"
  	depends on EVENTFD
  	select VDPA
+	select VHOST
  	help
  	  This kernel module can be loaded in host kernel to accelerate
  	  guest virtio devices with the vDPA-based backends.

@@ -78,5 +80,3 @@ config VHOST_CROSS_ENDIAN_LEGACY
  	  adds some overhead, it is disabled by default.
  
  	  If unsure, say "N".
-
-endif

But now CONFIG_VHOST is always "y", never "m".
Which I think will make it a built-in.
Didn't figure out why yet.