Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-23 Thread Gabriel L. Somlo 
On Tue, Feb 23, 2016 at 04:14:46PM +0200, Michael S. Tsirkin wrote:
> On Tue, Feb 23, 2016 at 08:47:00AM -0500, Gabriel L. Somlo wrote:
> > On Tue, Feb 23, 2016 at 07:07:36AM +0200, Michael S. Tsirkin wrote:
> > > On Mon, Feb 22, 2016 at 03:26:23PM -0500, Gabriel L. Somlo wrote:
> > > > On Mon, Feb 22, 2016 at 10:14:50PM +0200, Michael S. Tsirkin wrote:
> > > > > On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > > > > > > > +static void fw_cfg_io_cleanup(void)
> > > > > > > > +{
> > > > > > > > +   if (fw_cfg_is_mmio) {
> > > > > > > > +   iounmap(fw_cfg_dev_base);
> > > > > > > > +   release_mem_region(fw_cfg_p_base, 
> > > > > > > > fw_cfg_p_size);
> > > > > > > > +   } else {
> > > > > > > > +   ioport_unmap(fw_cfg_dev_base);
> > > > > > > > +   release_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > > > > +   }
> > > > > > > > +}
> > > > > > > > +
> > > > > > > > +/* arch-specific ctrl & data register offsets are not 
> > > > > > > > available in ACPI, DT */
> > > > > > > 
> > > > > > > So for all arches which support ACPI, I think this driver
> > > > > > > should just rely on ACPI.
> > > > > > 
> > > > > > There was a discussion about that a few versions ago, and IIRC the
> > > > > > conclusion was not to expect the firmware to contend for fw_cfg 
> > > > > > access
> > > > > > after the guest kernel boots:
> > > > > > 
> > > > > > https://lkml.org/lkml/2015/10/5/283
> > > > > > 
> > > > > 
> > > > > So it looks like NVDIMM at least wants to pass label data to guest -
> > > > > for which fw cfg might be a reasonable choice.
> > > > > 
> > > > > I suspect things changed - fw cfg used to be very slow but we now have
> > > > > DMA interface which makes it useful for a range of applications.
> > > 
> > > Comment on this? I'm really worried we'll release linux
> > > without a way to access fw cfg from aml.
> > > How about taking acpi lock around all accesses?
> > 
> > You mean something like this (haven't tried compiling it yet, so it
> > might be a bit more complicated, but just for the purpose of this
> > conversation):
> > 
> > diff --git a/drivers/firmware/qemu_fw_cfg.c
> > b/drivers/firmware/qemu_fw_cfg.c
> > index fedbff5..3462a2c 100644
> > --- a/drivers/firmware/qemu_fw_cfg.c
> > +++ b/drivers/firmware/qemu_fw_cfg.c
> > @@ -77,12 +77,18 @@ static inline u16 fw_cfg_sel_endianness(u16 key)
> >  static inline void fw_cfg_read_blob(u16 key,
> > void *buf, loff_t pos, size_t
> > count)
> >  {
> > +#ifdef CONFIG_ACPI
> > +   acpi_os_acquire_mutex(acpi_gbl_osi_mutex, ACPI_WAIT_FOREVER);
> > +#endif
> > mutex_lock(_cfg_dev_lock);
> > iowrite16(fw_cfg_sel_endianness(key), fw_cfg_reg_ctrl);
> > while (pos-- > 0)
> > ioread8(fw_cfg_reg_data);
> > ioread8_rep(fw_cfg_reg_data, buf, count);
> > mutex_unlock(_cfg_dev_lock);
> > +#ifdef CONFIG_ACPI
> > +   acpi_os_release_mutex(acpi_gbl_osi_mutex);
> > +#endif
> >  }
> >  
> >  /* clean up fw_cfg device i/o */
> 
> Fundamentally yes.
> 
> > I wouldn't particularly *mind* doing that, but I'd still like to hear
> > from other QEMU devs on whether it's really necessary.
> 
> It seems like a prudent thing to do IMHO, before this
> goes out to users.
> 
> [...]
> 
> On balance, I think locking ACPI solves most problems so
> if we just do that, I think what you did here is fine.

Only trouble is, acpi_gbl_osi_mutex seems to be "private" to the acpi
subsystem, and I'm not sure how well a patch to allow some random
module to lock/unlock ACPI at its convenience would be received...

So unless I'm missing something obvious (wouldn't be the first time),
I think we're back to where *if* we *have* to do this [*], providing an
AML blob-reader method in ACPI and punting to it from the guest-side
kernel module (via #ifdef CONFIG_ACPI) would be the less painful
alternative.

[*] that is, mutual exclusion between kernel and firmware regarding
fw_cfg is (back) on the table, for real this time...

It would be good to know that it's the new consensus among QEMU
folks, since I have a strong feeling I'd no longer be "Keeping It Simple"
by moving in this direction.

Thanks,
--Gabriel

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-23 Thread Michael S. Tsirkin 
On Tue, Feb 23, 2016 at 08:47:00AM -0500, Gabriel L. Somlo wrote:
> On Tue, Feb 23, 2016 at 07:07:36AM +0200, Michael S. Tsirkin wrote:
> > On Mon, Feb 22, 2016 at 03:26:23PM -0500, Gabriel L. Somlo wrote:
> > > On Mon, Feb 22, 2016 at 10:14:50PM +0200, Michael S. Tsirkin wrote:
> > > > On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > > > > > > +static void fw_cfg_io_cleanup(void)
> > > > > > > +{
> > > > > > > + if (fw_cfg_is_mmio) {
> > > > > > > + iounmap(fw_cfg_dev_base);
> > > > > > > + release_mem_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > > > + } else {
> > > > > > > + ioport_unmap(fw_cfg_dev_base);
> > > > > > > + release_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > > > + }
> > > > > > > +}
> > > > > > > +
> > > > > > > +/* arch-specific ctrl & data register offsets are not available 
> > > > > > > in ACPI, DT */
> > > > > > 
> > > > > > So for all arches which support ACPI, I think this driver
> > > > > > should just rely on ACPI.
> > > > > 
> > > > > There was a discussion about that a few versions ago, and IIRC the
> > > > > conclusion was not to expect the firmware to contend for fw_cfg access
> > > > > after the guest kernel boots:
> > > > > 
> > > > > https://lkml.org/lkml/2015/10/5/283
> > > > > 
> > > > 
> > > > So it looks like NVDIMM at least wants to pass label data to guest -
> > > > for which fw cfg might be a reasonable choice.
> > > > 
> > > > I suspect things changed - fw cfg used to be very slow but we now have
> > > > DMA interface which makes it useful for a range of applications.
> > 
> > Comment on this? I'm really worried we'll release linux
> > without a way to access fw cfg from aml.
> > How about taking acpi lock around all accesses?
> 
> You mean something like this (haven't tried compiling it yet, so it
> might be a bit more complicated, but just for the purpose of this
> conversation):
> 
> diff --git a/drivers/firmware/qemu_fw_cfg.c
> b/drivers/firmware/qemu_fw_cfg.c
> index fedbff5..3462a2c 100644
> --- a/drivers/firmware/qemu_fw_cfg.c
> +++ b/drivers/firmware/qemu_fw_cfg.c
> @@ -77,12 +77,18 @@ static inline u16 fw_cfg_sel_endianness(u16 key)
>  static inline void fw_cfg_read_blob(u16 key,
> void *buf, loff_t pos, size_t
> count)
>  {
> +#ifdef CONFIG_ACPI
> +   acpi_os_acquire_mutex(acpi_gbl_osi_mutex, ACPI_WAIT_FOREVER);
> +#endif
> mutex_lock(_cfg_dev_lock);
> iowrite16(fw_cfg_sel_endianness(key), fw_cfg_reg_ctrl);
> while (pos-- > 0)
> ioread8(fw_cfg_reg_data);
> ioread8_rep(fw_cfg_reg_data, buf, count);
> mutex_unlock(_cfg_dev_lock);
> +#ifdef CONFIG_ACPI
> +   acpi_os_release_mutex(acpi_gbl_osi_mutex);
> +#endif
>  }
>  
>  /* clean up fw_cfg device i/o */

Fundamentally yes.

> I wouldn't particularly *mind* doing that, but I'd still like to hear
> from other QEMU devs on whether it's really necessary.

It seems like a prudent thing to do IMHO, before this
goes out to users.

> > > > > (I even had a prototype version doing what you suggested, but per the 
> > > > > above
> > > > > reference decided to drop it -- which IMHO is for the better, since 
> > > > > otherwise
> > > > > I'd have had to ifdef between ACPI and non-ACPI versions of the 
> > > > > driver --
> > > > > see https://lkml.org/lkml/2015/11/4/534 )
> > > > 
> > > > I'm not sure why you have these ifdefs - they are on the host, are they
> > > > not?
> > > 
> > > Think of those as "pseudocode" ifdefs, they're there to distinguish
> > > between AML that would be generated on MMIO vs. IOPORT systems
> > > (specifically, arm vs. x86, respectively)
> > > 
> > > Some of the AML is the same, but obviously the _CRS, and
> > > OperationRegion + Field are different, and I wanted to point that out
> > > somehow :)
> > > 
> > > Cheers,
> > > --Gabriel
> > 
> > You can do ifs as well.
> 
> Yeah, but the AML is generated from arch-specific locations in QEMU,
> so we'd be doing MMIO-only from e.g. hw/arm/virt-acpi-build.c, and
> IOPORT-only from hw/i386/acpi-build.c, etc. I wouldnt need to write a
> generic AML blob with 'if' statements and insert it the same way on
> all architectures, or would I ? Not sure what the best practice would
> be for that :)

Just regular C, put common code in a common function.

> Speaking of AML, if we were to implement a "RDBL" (read-blob) method
> for fw_cfg in AML, and call it from the guest-side kernel module,
> we'll never be able to make it use DMA on ACPI systems. The way
> fw_cfg_read_blob is written now, we could patch that in at some later
> point. So that's an argument in favor of *at most* wrapping
> acpi_os_acquire_mutex() around the current fw_cfg_read_blob, rather
> than including an acpi-specific version implemented on top of an
> AML call.
> 
> Thanks,
> --Gabriel

On balance, I think locking ACPI solves most problems so
if we just do that, I think what you did here is fine.

-- 
MST

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-23 Thread Gabriel L. Somlo 
On Tue, Feb 23, 2016 at 07:07:36AM +0200, Michael S. Tsirkin wrote:
> On Mon, Feb 22, 2016 at 03:26:23PM -0500, Gabriel L. Somlo wrote:
> > On Mon, Feb 22, 2016 at 10:14:50PM +0200, Michael S. Tsirkin wrote:
> > > On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > > > > > +static void fw_cfg_io_cleanup(void)
> > > > > > +{
> > > > > > +   if (fw_cfg_is_mmio) {
> > > > > > +   iounmap(fw_cfg_dev_base);
> > > > > > +   release_mem_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > > +   } else {
> > > > > > +   ioport_unmap(fw_cfg_dev_base);
> > > > > > +   release_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > > +   }
> > > > > > +}
> > > > > > +
> > > > > > +/* arch-specific ctrl & data register offsets are not available in 
> > > > > > ACPI, DT */
> > > > > 
> > > > > So for all arches which support ACPI, I think this driver
> > > > > should just rely on ACPI.
> > > > 
> > > > There was a discussion about that a few versions ago, and IIRC the
> > > > conclusion was not to expect the firmware to contend for fw_cfg access
> > > > after the guest kernel boots:
> > > > 
> > > > https://lkml.org/lkml/2015/10/5/283
> > > > 
> > > 
> > > So it looks like NVDIMM at least wants to pass label data to guest -
> > > for which fw cfg might be a reasonable choice.
> > > 
> > > I suspect things changed - fw cfg used to be very slow but we now have
> > > DMA interface which makes it useful for a range of applications.
> 
> Comment on this? I'm really worried we'll release linux
> without a way to access fw cfg from aml.
> How about taking acpi lock around all accesses?

You mean something like this (haven't tried compiling it yet, so it
might be a bit more complicated, but just for the purpose of this
conversation):

diff --git a/drivers/firmware/qemu_fw_cfg.c
b/drivers/firmware/qemu_fw_cfg.c
index fedbff5..3462a2c 100644
--- a/drivers/firmware/qemu_fw_cfg.c
+++ b/drivers/firmware/qemu_fw_cfg.c
@@ -77,12 +77,18 @@ static inline u16 fw_cfg_sel_endianness(u16 key)
 static inline void fw_cfg_read_blob(u16 key,
void *buf, loff_t pos, size_t
count)
 {
+#ifdef CONFIG_ACPI
+   acpi_os_acquire_mutex(acpi_gbl_osi_mutex, ACPI_WAIT_FOREVER);
+#endif
mutex_lock(_cfg_dev_lock);
iowrite16(fw_cfg_sel_endianness(key), fw_cfg_reg_ctrl);
while (pos-- > 0)
ioread8(fw_cfg_reg_data);
ioread8_rep(fw_cfg_reg_data, buf, count);
mutex_unlock(_cfg_dev_lock);
+#ifdef CONFIG_ACPI
+   acpi_os_release_mutex(acpi_gbl_osi_mutex);
+#endif
 }
 
 /* clean up fw_cfg device i/o */

I wouldn't particularly *mind* doing that, but I'd still like to hear
from other QEMU devs on whether it's really necessary.

> > > > (I even had a prototype version doing what you suggested, but per the 
> > > > above
> > > > reference decided to drop it -- which IMHO is for the better, since 
> > > > otherwise
> > > > I'd have had to ifdef between ACPI and non-ACPI versions of the driver 
> > > > --
> > > > see https://lkml.org/lkml/2015/11/4/534 )
> > > 
> > > I'm not sure why you have these ifdefs - they are on the host, are they
> > > not?
> > 
> > Think of those as "pseudocode" ifdefs, they're there to distinguish
> > between AML that would be generated on MMIO vs. IOPORT systems
> > (specifically, arm vs. x86, respectively)
> > 
> > Some of the AML is the same, but obviously the _CRS, and
> > OperationRegion + Field are different, and I wanted to point that out
> > somehow :)
> > 
> > Cheers,
> > --Gabriel
> 
> You can do ifs as well.

Yeah, but the AML is generated from arch-specific locations in QEMU,
so we'd be doing MMIO-only from e.g. hw/arm/virt-acpi-build.c, and
IOPORT-only from hw/i386/acpi-build.c, etc. I wouldnt need to write a
generic AML blob with 'if' statements and insert it the same way on
all architectures, or would I ? Not sure what the best practice would
be for that :)

Speaking of AML, if we were to implement a "RDBL" (read-blob) method
for fw_cfg in AML, and call it from the guest-side kernel module,
we'll never be able to make it use DMA on ACPI systems. The way
fw_cfg_read_blob is written now, we could patch that in at some later
point. So that's an argument in favor of *at most* wrapping
acpi_os_acquire_mutex() around the current fw_cfg_read_blob, rather
than including an acpi-specific version implemented on top of an
AML call.

Thanks,
--Gabriel

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-22 Thread Michael S. Tsirkin 
On Mon, Feb 22, 2016 at 03:26:23PM -0500, Gabriel L. Somlo wrote:
> On Mon, Feb 22, 2016 at 10:14:50PM +0200, Michael S. Tsirkin wrote:
> > On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > > > > +static void fw_cfg_io_cleanup(void)
> > > > > +{
> > > > > + if (fw_cfg_is_mmio) {
> > > > > + iounmap(fw_cfg_dev_base);
> > > > > + release_mem_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > + } else {
> > > > > + ioport_unmap(fw_cfg_dev_base);
> > > > > + release_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > > + }
> > > > > +}
> > > > > +
> > > > > +/* arch-specific ctrl & data register offsets are not available in 
> > > > > ACPI, DT */
> > > > 
> > > > So for all arches which support ACPI, I think this driver
> > > > should just rely on ACPI.
> > > 
> > > There was a discussion about that a few versions ago, and IIRC the
> > > conclusion was not to expect the firmware to contend for fw_cfg access
> > > after the guest kernel boots:
> > > 
> > > https://lkml.org/lkml/2015/10/5/283
> > > 
> > 
> > So it looks like NVDIMM at least wants to pass label data to guest -
> > for which fw cfg might be a reasonable choice.
> > 
> > I suspect things changed - fw cfg used to be very slow but we now have
> > DMA interface which makes it useful for a range of applications.

Comment on this? I'm really worried we'll release linux
without a way to access fw cfg from aml.
How about taking acpi lock around all accesses?

> > > (I even had a prototype version doing what you suggested, but per the 
> > > above
> > > reference decided to drop it -- which IMHO is for the better, since 
> > > otherwise
> > > I'd have had to ifdef between ACPI and non-ACPI versions of the driver --
> > > see https://lkml.org/lkml/2015/11/4/534 )
> > 
> > I'm not sure why you have these ifdefs - they are on the host, are they
> > not?
> 
> Think of those as "pseudocode" ifdefs, they're there to distinguish
> between AML that would be generated on MMIO vs. IOPORT systems
> (specifically, arm vs. x86, respectively)
> 
> Some of the AML is the same, but obviously the _CRS, and
> OperationRegion + Field are different, and I wanted to point that out
> somehow :)
> 
> Cheers,
> --Gabriel

You can do ifs as well.

-- 
MST

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-22 Thread Gabriel L. Somlo 
On Mon, Feb 22, 2016 at 10:14:50PM +0200, Michael S. Tsirkin wrote:
> On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > > > +static void fw_cfg_io_cleanup(void)
> > > > +{
> > > > +   if (fw_cfg_is_mmio) {
> > > > +   iounmap(fw_cfg_dev_base);
> > > > +   release_mem_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > +   } else {
> > > > +   ioport_unmap(fw_cfg_dev_base);
> > > > +   release_region(fw_cfg_p_base, fw_cfg_p_size);
> > > > +   }
> > > > +}
> > > > +
> > > > +/* arch-specific ctrl & data register offsets are not available in 
> > > > ACPI, DT */
> > > 
> > > So for all arches which support ACPI, I think this driver
> > > should just rely on ACPI.
> > 
> > There was a discussion about that a few versions ago, and IIRC the
> > conclusion was not to expect the firmware to contend for fw_cfg access
> > after the guest kernel boots:
> > 
> > https://lkml.org/lkml/2015/10/5/283
> > 
> 
> So it looks like NVDIMM at least wants to pass label data to guest -
> for which fw cfg might be a reasonable choice.
> 
> I suspect things changed - fw cfg used to be very slow but we now have
> DMA interface which makes it useful for a range of applications.
> 
> > (I even had a prototype version doing what you suggested, but per the above
> > reference decided to drop it -- which IMHO is for the better, since 
> > otherwise
> > I'd have had to ifdef between ACPI and non-ACPI versions of the driver --
> > see https://lkml.org/lkml/2015/11/4/534 )
> 
> I'm not sure why you have these ifdefs - they are on the host, are they
> not?

Think of those as "pseudocode" ifdefs, they're there to distinguish
between AML that would be generated on MMIO vs. IOPORT systems
(specifically, arm vs. x86, respectively)

Some of the AML is the same, but obviously the _CRS, and
OperationRegion + Field are different, and I wanted to point that out
somehow :)

Cheers,
--Gabriel

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-22 Thread Michael S. Tsirkin 
On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > > +static void fw_cfg_io_cleanup(void)
> > > +{
> > > + if (fw_cfg_is_mmio) {
> > > + iounmap(fw_cfg_dev_base);
> > > + release_mem_region(fw_cfg_p_base, fw_cfg_p_size);
> > > + } else {
> > > + ioport_unmap(fw_cfg_dev_base);
> > > + release_region(fw_cfg_p_base, fw_cfg_p_size);
> > > + }
> > > +}
> > > +
> > > +/* arch-specific ctrl & data register offsets are not available in ACPI, 
> > > DT */
> > 
> > So for all arches which support ACPI, I think this driver
> > should just rely on ACPI.
> 
> There was a discussion about that a few versions ago, and IIRC the
> conclusion was not to expect the firmware to contend for fw_cfg access
> after the guest kernel boots:
> 
> https://lkml.org/lkml/2015/10/5/283
> 

So it looks like NVDIMM at least wants to pass label data to guest -
for which fw cfg might be a reasonable choice.

I suspect things changed - fw cfg used to be very slow but we now have
DMA interface which makes it useful for a range of applications.

> (I even had a prototype version doing what you suggested, but per the above
> reference decided to drop it -- which IMHO is for the better, since otherwise
> I'd have had to ifdef between ACPI and non-ACPI versions of the driver --
> see https://lkml.org/lkml/2015/11/4/534 )

I'm not sure why you have these ifdefs - they are on the host, are they
not?

-- 
MST

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-21 Thread Gabriel L. Somlo 
On Sun, Feb 21, 2016 at 03:10:30PM +0200, Michael S. Tsirkin wrote:
> On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > 
> > > 
> > > > +#if !(defined(FW_CFG_CTRL_OFF) && defined(FW_CTRL_DATA_OFF))
> > > > +# if (defined(CONFIG_ARM) || defined(CONFIG_ARM64))
> > > > +#  define FW_CFG_CTRL_OFF 0x08
> > > > +#  define FW_CFG_DATA_OFF 0x00
> > > > +# elif (defined(CONFIG_PPC_PMAC) || defined(CONFIG_SPARC32)) /* 
> > > > ppc/mac,sun4m */
> > > > +#  define FW_CFG_CTRL_OFF 0x00
> > > > +#  define FW_CFG_DATA_OFF 0x02
> > > > +# elif (defined(CONFIG_X86) || defined(CONFIG_SPARC64)) /* x86, sun4u 
> > > > */
> > > > +#  define FW_CFG_CTRL_OFF 0x00
> > > > +#  define FW_CFG_DATA_OFF 0x01
> > > > +# else
> > > > +#  warning "QEMU FW_CFG may not be available on this architecture!"
> > > > +#  define FW_CFG_CTRL_OFF 0x00
> > > > +#  define FW_CFG_DATA_OFF 0x01
> > > 
> > > Better not try hacks like this, they are hard
> > > to support down the road. Please only list what is tested and
> > > actually exposed by QEMU.
> > 
> > I was looking for a standard way to advertise register offsets within
> > the ioport or mmio region assigned to fw_cfg, but right now the answer
> > is "there isn't one", and "register offsets are an arch specific
> > detail". As such, the only reasonable way I saw was to copy the same
> > values used in the QEMU source.
> > 
> > See also:
> > https://lists.gnu.org/archive/html/qemu-devel/2015-11/msg05037.html
> >  
> > Thanks much,
> > --Gabriel
> 
> My point is you don't know what will qemu do on these
> other arches which do not at the moment have fw cfg.
> So don't try to guess!

Oh, you mean for the "else". I originally wanted to be able to compile
this on any architecture and wanted some dummy defaults I could
override on the command line. But now we're already restricting this
to known architectures only, so I'll send a patch turning the warning
into an error, and removing the #defines for the "else" branch above.

Sorry I misunderstood you the first time around :)

Thanks,
--Gabriel

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-21 Thread Michael S. Tsirkin 
On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> > So for all arches which support ACPI, I think this driver
> > should just rely on ACPI.
> 
> There was a discussion about that a few versions ago, and IIRC the
> conclusion was not to expect the firmware to contend for fw_cfg access
> after the guest kernel boots:
> 
> https://lkml.org/lkml/2015/10/5/283

Interesting. Igor wanted to do this again recently ...
I'll think it over and get back to you/list on this.

> (I even had a prototype version doing what you suggested, but per the above
> reference decided to drop it -- which IMHO is for the better, since otherwise
> I'd have had to ifdef between ACPI and non-ACPI versions of the driver --
> see https://lkml.org/lkml/2015/11/4/534 )

I'm not sure I get it - won't you only ifdef the accessor function?

-- 
MST

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-21 Thread Michael S. Tsirkin 
On Sun, Feb 21, 2016 at 08:06:17AM -0500, Gabriel L. Somlo wrote:
> 
> > 
> > > +#if !(defined(FW_CFG_CTRL_OFF) && defined(FW_CTRL_DATA_OFF))
> > > +# if (defined(CONFIG_ARM) || defined(CONFIG_ARM64))
> > > +#  define FW_CFG_CTRL_OFF 0x08
> > > +#  define FW_CFG_DATA_OFF 0x00
> > > +# elif (defined(CONFIG_PPC_PMAC) || defined(CONFIG_SPARC32)) /* 
> > > ppc/mac,sun4m */
> > > +#  define FW_CFG_CTRL_OFF 0x00
> > > +#  define FW_CFG_DATA_OFF 0x02
> > > +# elif (defined(CONFIG_X86) || defined(CONFIG_SPARC64)) /* x86, sun4u */
> > > +#  define FW_CFG_CTRL_OFF 0x00
> > > +#  define FW_CFG_DATA_OFF 0x01
> > > +# else
> > > +#  warning "QEMU FW_CFG may not be available on this architecture!"
> > > +#  define FW_CFG_CTRL_OFF 0x00
> > > +#  define FW_CFG_DATA_OFF 0x01
> > 
> > Better not try hacks like this, they are hard
> > to support down the road. Please only list what is tested and
> > actually exposed by QEMU.
> 
> I was looking for a standard way to advertise register offsets within
> the ioport or mmio region assigned to fw_cfg, but right now the answer
> is "there isn't one", and "register offsets are an arch specific
> detail". As such, the only reasonable way I saw was to copy the same
> values used in the QEMU source.
> 
> See also:
> https://lists.gnu.org/archive/html/qemu-devel/2015-11/msg05037.html
>  
> Thanks much,
> --Gabriel

My point is you don't know what will qemu do on these
other arches which do not at the moment have fw cfg.
So don't try to guess!

-- 
MST

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-21 Thread Gabriel L. Somlo 
On Sun, Feb 21, 2016 at 10:30:26AM +0200, Michael S. Tsirkin wrote:
> On Thu, Jan 28, 2016 at 09:23:11AM -0500, Gabriel L. Somlo wrote:
> > From: Gabriel Somlo 
> > 
> > Make fw_cfg entries of type "file" available via sysfs. Entries
> > are listed under /sys/firmware/qemu_fw_cfg/by_key, in folders
> > named after each entry's selector key. Filename, selector value,
> > and size read-only attributes are included for each entry. Also,
> > a "raw" attribute allows retrieval of the full binary content of
> > each entry.
> > 
> > The fw_cfg device can be instantiated automatically from ACPI or
> > the Device Tree, or manually by using a kernel module (or command
> > line) parameter, with a syntax outlined in the documentation file.
> > 
> > Signed-off-by: Gabriel Somlo 
> > ---
> >  .../ABI/testing/sysfs-firmware-qemu_fw_cfg |  58 ++
> >  drivers/firmware/Kconfig   |  19 +
> >  drivers/firmware/Makefile  |   1 +
> >  drivers/firmware/qemu_fw_cfg.c | 648 
> > +
> >  4 files changed, 726 insertions(+)
> >  create mode 100644 Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
> >  create mode 100644 drivers/firmware/qemu_fw_cfg.c
> > 
> > diff --git a/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg 
> > b/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
> > new file mode 100644
> > index 000..e9e58d4
> > --- /dev/null
> > +++ b/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
> > @@ -0,0 +1,58 @@
> > +What:  /sys/firmware/qemu_fw_cfg/
> > +Date:  August 2015
> > +Contact:   Gabriel Somlo 
> > +Description:
> > +   Several different architectures supported by QEMU (x86, arm,
> > +   sun4*, ppc/mac) are provisioned with a firmware configuration
> > +   (fw_cfg) device, originally intended as a way for the host to
> > +   provide configuration data to the guest firmware. Starting
> > +   with QEMU v2.4, arbitrary fw_cfg file entries may be specified
> > +   by the user on the command line, which makes fw_cfg additionally
> > +   useful as an out-of-band, asynchronous mechanism for providing
> > +   configuration data to the guest userspace.
> > +
> > +   The authoritative guest-side hardware interface documentation
> > +   to the fw_cfg device can be found in "docs/specs/fw_cfg.txt"
> > +   in the QEMU source tree.
> > +
> > +   === SysFS fw_cfg Interface ===
> > +
> > +   The fw_cfg sysfs interface described in this document is only
> > +   intended to display discoverable blobs (i.e., those registered
> > +   with the file directory), as there is no way to determine the
> > +   presence or size of "legacy" blobs (with selector keys between
> > +   0x0002 and 0x0018) programmatically.
> > +
> > +   All fw_cfg information is shown under:
> > +
> > +   /sys/firmware/qemu_fw_cfg/
> > +
> > +   The only legacy blob displayed is the fw_cfg device revision:
> > +
> > +   /sys/firmware/qemu_fw_cfg/rev
> > +
> > +   --- Discoverable fw_cfg blobs by selector key ---
> > +
> > +   All discoverable blobs listed in the fw_cfg file directory are
> > +   displayed as entries named after their unique selector key
> > +   value, e.g.:
> > +
> > +   /sys/firmware/qemu_fw_cfg/by_key/32
> > +   /sys/firmware/qemu_fw_cfg/by_key/33
> > +   /sys/firmware/qemu_fw_cfg/by_key/34
> > +   ...
> > +
> > +   Each such fw_cfg sysfs entry has the following values exported
> > +   as attributes:
> > +
> > +   name: The 56-byte nul-terminated ASCII string used as the
> > + blob's 'file name' in the fw_cfg directory.
> > +   size: The length of the blob, as given in the fw_cfg
> > + directory.
> > +   key : The value of the blob's selector key as given in the
> > + fw_cfg directory. This value is the same as used in
> > + the parent directory name.
> > +   raw : The raw bytes of the blob, obtained by selecting the
> > + entry via the control register, and reading a number
> > + of bytes equal to the blob size from the data
> > + register.
> > diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
> > index 49a3a11..5130f74 100644
> > --- a/drivers/firmware/Kconfig
> > +++ b/drivers/firmware/Kconfig
> > @@ -161,6 +161,25 @@ config RASPBERRYPI_FIRMWARE
> >   This option enables support for communicating with the firmware on the
> >   Raspberry Pi.
> >  
> > +config FW_CFG_SYSFS
> > +   tristate "QEMU fw_cfg device support in sysfs"
> > +   depends on SYSFS && (ARM || ARM64 ||

Re: [Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-02-21 Thread Michael S. Tsirkin 
On Thu, Jan 28, 2016 at 09:23:11AM -0500, Gabriel L. Somlo wrote:
> From: Gabriel Somlo 
> 
> Make fw_cfg entries of type "file" available via sysfs. Entries
> are listed under /sys/firmware/qemu_fw_cfg/by_key, in folders
> named after each entry's selector key. Filename, selector value,
> and size read-only attributes are included for each entry. Also,
> a "raw" attribute allows retrieval of the full binary content of
> each entry.
> 
> The fw_cfg device can be instantiated automatically from ACPI or
> the Device Tree, or manually by using a kernel module (or command
> line) parameter, with a syntax outlined in the documentation file.
> 
> Signed-off-by: Gabriel Somlo 
> ---
>  .../ABI/testing/sysfs-firmware-qemu_fw_cfg |  58 ++
>  drivers/firmware/Kconfig   |  19 +
>  drivers/firmware/Makefile  |   1 +
>  drivers/firmware/qemu_fw_cfg.c | 648 
> +
>  4 files changed, 726 insertions(+)
>  create mode 100644 Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
>  create mode 100644 drivers/firmware/qemu_fw_cfg.c
> 
> diff --git a/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg 
> b/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
> new file mode 100644
> index 000..e9e58d4
> --- /dev/null
> +++ b/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
> @@ -0,0 +1,58 @@
> +What:/sys/firmware/qemu_fw_cfg/
> +Date:August 2015
> +Contact: Gabriel Somlo 
> +Description:
> + Several different architectures supported by QEMU (x86, arm,
> + sun4*, ppc/mac) are provisioned with a firmware configuration
> + (fw_cfg) device, originally intended as a way for the host to
> + provide configuration data to the guest firmware. Starting
> + with QEMU v2.4, arbitrary fw_cfg file entries may be specified
> + by the user on the command line, which makes fw_cfg additionally
> + useful as an out-of-band, asynchronous mechanism for providing
> + configuration data to the guest userspace.
> +
> + The authoritative guest-side hardware interface documentation
> + to the fw_cfg device can be found in "docs/specs/fw_cfg.txt"
> + in the QEMU source tree.
> +
> + === SysFS fw_cfg Interface ===
> +
> + The fw_cfg sysfs interface described in this document is only
> + intended to display discoverable blobs (i.e., those registered
> + with the file directory), as there is no way to determine the
> + presence or size of "legacy" blobs (with selector keys between
> + 0x0002 and 0x0018) programmatically.
> +
> + All fw_cfg information is shown under:
> +
> + /sys/firmware/qemu_fw_cfg/
> +
> + The only legacy blob displayed is the fw_cfg device revision:
> +
> + /sys/firmware/qemu_fw_cfg/rev
> +
> + --- Discoverable fw_cfg blobs by selector key ---
> +
> + All discoverable blobs listed in the fw_cfg file directory are
> + displayed as entries named after their unique selector key
> + value, e.g.:
> +
> + /sys/firmware/qemu_fw_cfg/by_key/32
> + /sys/firmware/qemu_fw_cfg/by_key/33
> + /sys/firmware/qemu_fw_cfg/by_key/34
> + ...
> +
> + Each such fw_cfg sysfs entry has the following values exported
> + as attributes:
> +
> + name: The 56-byte nul-terminated ASCII string used as the
> +   blob's 'file name' in the fw_cfg directory.
> + size: The length of the blob, as given in the fw_cfg
> +   directory.
> + key : The value of the blob's selector key as given in the
> +   fw_cfg directory. This value is the same as used in
> +   the parent directory name.
> + raw : The raw bytes of the blob, obtained by selecting the
> +   entry via the control register, and reading a number
> +   of bytes equal to the blob size from the data
> +   register.
> diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
> index 49a3a11..5130f74 100644
> --- a/drivers/firmware/Kconfig
> +++ b/drivers/firmware/Kconfig
> @@ -161,6 +161,25 @@ config RASPBERRYPI_FIRMWARE
> This option enables support for communicating with the firmware on the
> Raspberry Pi.
>  
> +config FW_CFG_SYSFS
> + tristate "QEMU fw_cfg device support in sysfs"
> + depends on SYSFS && (ARM || ARM64 || PPC_PMAC || SPARC || X86)
> + default n
> + help
> +   Say Y or M here to enable the exporting of the QEMU firmware
> +   configuration (fw_cfg) file

[Qemu-devel] [PATCH v8 1/4] firmware: introduce sysfs driver for QEMU's fw_cfg device

2016-01-28 Thread Gabriel L. Somlo 
From: Gabriel Somlo 

Make fw_cfg entries of type "file" available via sysfs. Entries
are listed under /sys/firmware/qemu_fw_cfg/by_key, in folders
named after each entry's selector key. Filename, selector value,
and size read-only attributes are included for each entry. Also,
a "raw" attribute allows retrieval of the full binary content of
each entry.

The fw_cfg device can be instantiated automatically from ACPI or
the Device Tree, or manually by using a kernel module (or command
line) parameter, with a syntax outlined in the documentation file.

Signed-off-by: Gabriel Somlo 
---
 .../ABI/testing/sysfs-firmware-qemu_fw_cfg |  58 ++
 drivers/firmware/Kconfig   |  19 +
 drivers/firmware/Makefile  |   1 +
 drivers/firmware/qemu_fw_cfg.c | 648 +
 4 files changed, 726 insertions(+)
 create mode 100644 Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
 create mode 100644 drivers/firmware/qemu_fw_cfg.c

diff --git a/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg 
b/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
new file mode 100644
index 000..e9e58d4
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-firmware-qemu_fw_cfg
@@ -0,0 +1,58 @@
+What:  /sys/firmware/qemu_fw_cfg/
+Date:  August 2015
+Contact:   Gabriel Somlo 
+Description:
+   Several different architectures supported by QEMU (x86, arm,
+   sun4*, ppc/mac) are provisioned with a firmware configuration
+   (fw_cfg) device, originally intended as a way for the host to
+   provide configuration data to the guest firmware. Starting
+   with QEMU v2.4, arbitrary fw_cfg file entries may be specified
+   by the user on the command line, which makes fw_cfg additionally
+   useful as an out-of-band, asynchronous mechanism for providing
+   configuration data to the guest userspace.
+
+   The authoritative guest-side hardware interface documentation
+   to the fw_cfg device can be found in "docs/specs/fw_cfg.txt"
+   in the QEMU source tree.
+
+   === SysFS fw_cfg Interface ===
+
+   The fw_cfg sysfs interface described in this document is only
+   intended to display discoverable blobs (i.e., those registered
+   with the file directory), as there is no way to determine the
+   presence or size of "legacy" blobs (with selector keys between
+   0x0002 and 0x0018) programmatically.
+
+   All fw_cfg information is shown under:
+
+   /sys/firmware/qemu_fw_cfg/
+
+   The only legacy blob displayed is the fw_cfg device revision:
+
+   /sys/firmware/qemu_fw_cfg/rev
+
+   --- Discoverable fw_cfg blobs by selector key ---
+
+   All discoverable blobs listed in the fw_cfg file directory are
+   displayed as entries named after their unique selector key
+   value, e.g.:
+
+   /sys/firmware/qemu_fw_cfg/by_key/32
+   /sys/firmware/qemu_fw_cfg/by_key/33
+   /sys/firmware/qemu_fw_cfg/by_key/34
+   ...
+
+   Each such fw_cfg sysfs entry has the following values exported
+   as attributes:
+
+   name: The 56-byte nul-terminated ASCII string used as the
+ blob's 'file name' in the fw_cfg directory.
+   size: The length of the blob, as given in the fw_cfg
+ directory.
+   key : The value of the blob's selector key as given in the
+ fw_cfg directory. This value is the same as used in
+ the parent directory name.
+   raw : The raw bytes of the blob, obtained by selecting the
+ entry via the control register, and reading a number
+ of bytes equal to the blob size from the data
+ register.
diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
index 49a3a11..5130f74 100644
--- a/drivers/firmware/Kconfig
+++ b/drivers/firmware/Kconfig
@@ -161,6 +161,25 @@ config RASPBERRYPI_FIRMWARE
  This option enables support for communicating with the firmware on the
  Raspberry Pi.
 
+config FW_CFG_SYSFS
+   tristate "QEMU fw_cfg device support in sysfs"
+   depends on SYSFS && (ARM || ARM64 || PPC_PMAC || SPARC || X86)
+   default n
+   help
+ Say Y or M here to enable the exporting of the QEMU firmware
+ configuration (fw_cfg) file entries via sysfs. Entries are
+ found under /sys/firmware/fw_cfg when this option is enabled
+ and loaded.
+
+config FW_CFG_SYSFS_CMDLINE
+   bool "QEMU fw_cfg device