Re: [RFC PATCH 09/14] cxl/mem: Add basic IOCTL interface

From: Dan Williams
Date: Tue Dec 08 2020 - 22:34:47 EST


On Tue, Dec 8, 2020 at 6:13 PM Ben Widawsky <ben.widawsky@xxxxxxxxx> wrote:
>
> On 20-12-08 17:37:50, Dan Williams wrote:
> > On Tue, Dec 8, 2020 at 4:24 PM Ben Widawsky <ben.widawsky@xxxxxxxxx> wrote:
> > >
> > > Add a straightforward IOCTL that provides a mechanism for userspace to
> > > query the supported memory device commands.
> > >
> > > Memory device commands are specified in 8.2.9 of the CXL 2.0
> > > specification. They are submitted through a mailbox mechanism specified
> > > in 8.2.8.4.
> > >
> > > Signed-off-by: Ben Widawsky <ben.widawsky@xxxxxxxxx>
> > >
> > > ---
> > >
> > > I did attempt to use the same struct for querying commands as well as
> > > sending commands (upcoming patch). The number of unused fields between
> > > the two made for a bad fit IMO.
> > >
> > > Signed-off-by: Ben Widawsky <ben.widawsky@xxxxxxxxx>
> > > ---
> > > Documentation/cxl/memory-devices.rst | 9 +++
> > > drivers/cxl/mem.c | 89 +++++++++++++++++++++++
> > > include/uapi/linux/cxl_mem.h | 102 +++++++++++++++++++++++++++
> > > 3 files changed, 200 insertions(+)
> > > create mode 100644 include/uapi/linux/cxl_mem.h
> > >
> > > diff --git a/Documentation/cxl/memory-devices.rst b/Documentation/cxl/memory-devices.rst
> > > index 5f723c25382b..ec54674b3822 100644
> > > --- a/Documentation/cxl/memory-devices.rst
> > > +++ b/Documentation/cxl/memory-devices.rst
> > > @@ -32,6 +32,15 @@ CXL Memory Device
> > > .. kernel-doc:: drivers/cxl/mem.c
> > > :internal:
> > >
> > > +CXL IOCTL Interface
> > > +-------------------
> > > +
> > > +.. kernel-doc:: include/uapi/linux/cxl_mem.h
> > > + :doc: UAPI
> > > +
> > > +.. kernel-doc:: include/uapi/linux/cxl_mem.h
> > > + :internal:
> > > +
> > > External Interfaces
> > > ===================
> > >
> > > diff --git a/drivers/cxl/mem.c b/drivers/cxl/mem.c
> > > index bb6ea58f6c7b..2c4aadcea0e4 100644
> > > --- a/drivers/cxl/mem.c
> > > +++ b/drivers/cxl/mem.c
> > > @@ -7,6 +7,7 @@
> > > #include <linux/idr.h>
> > > #include <linux/pci.h>
> > > #include <linux/io.h>
> > > +#include <uapi/linux/cxl_mem.h>
> > > #include "acpi.h"
> > > #include "pci.h"
> > > #include "cxl.h"
> > > @@ -73,6 +74,49 @@ static DEFINE_IDR(cxl_mem_idr);
> > > /* protect cxl_mem_idr allocations */
> > > static DEFINE_MUTEX(cxl_memdev_lock);
> > >
> > > +/*
> > > + * This table defines the supported mailboxes commands for the driver. The id is
> > > + * ordinal and thus gaps in this table aren't allowed. This table is made up of
> > > + * a UAPI structure. Non-negative values in the table will be validated against
> > > + * the user's input. For example, if size_in is 0, and the user passed in 1, it
> > > + * is an error.
> > > + */
> > > +#define CXL_CMD(_id, _flags, sin, sout, _name, _enable, op) \
> > > + { \
> > > + { .id = CXL_MEM_COMMAND_ID_##_id, \
> > > + .flags = CXL_MEM_COMMAND_FLAG_##_flags, \
> > > + .size_in = sin, \
> > > + .size_out = sout, \
> > > + .name = _name }, \
> > > + .enable = _enable, .opcode = op \
> > > + }
> >
> > Seems the ordinality requirement could be dropped if the definition was:
> >
> > #define CXL_CMD(_id, _flags, sin, sout, _name, _enable, op) \
> > [CXL_MEM_COMMAND_ID_##_id] = {
> > \
> > { .id = CXL_MEM_COMMAND_ID_##_id, \
> > ...
> >
> > Then command 0 and 42 could be defined out of order in the table.
> > Especially if we need to config-disable or deprecate commands, I think
> > it would be useful if this table was tolerant to being sparse.
> >
>
> How sparse are we talking? The current form does support sparseness, but
> obviously gets quite large if the ID numbering is similar to random
> distribution.

"Sparse" may have been the wrong word to use. I was implying sparse
enough that if I add command N+1 I don't need to be careful where I
put it in mem_commands, but still be able to rely on lookups into
mem_commands being indexed by the command-id.

> I think if we do see this being more like random distribution, it can be
> supported, but I think it adds a decent amount of complexity for what I see as
> not much reward - unless you know of a fairly simple way to create this data
> structure with full sparse ID support?

I'm expecting the command distribution to be mostly uniform, it's more
of the lookup property that I think would be useful especially for the
dynamic case of walking mem_commands to update it relative to what the
hardware supports or other metadata. Speaking of which I think @enable
should be turned into @flags of which 'enable' is one, in case we want
to define more flags in the future.

>
> > > +
> > > +/**
> > > + * struct cxl_mem_command - Driver representation of a memory device command
> > > + * @info: Command information as it exists for the UAPI
> > > + * @opcode: The actual bits used for the mailbox protocol
> > > + * @enable: Whether the command is enabled. The driver may support a large set
> > > + * of commands that may not be enabled. The primary reason a command
> > > + * would not be enabled is for commands that are specified as optional
> > > + * and the hardware doesn't support the command.
> > > + *
> > > + * The cxl_mem_command is the driver's internal representation of commands that
> > > + * are supported by the driver. Some of these commands may not be supported by
> > > + * the hardware (!@enable). The driver will use @info to validate the fields
> > > + * passed in by the user then submit the @opcode to the hardware.
> > > + *
> > > + * See struct cxl_command_info.
> > > + */
> > > +struct cxl_mem_command {
> > > + const struct cxl_command_info info;
> > > + const u16 opcode;
> > > + bool enable;
> > > +};
> > > +
> > > +static struct cxl_mem_command mem_commands[] = {
> > > + CXL_CMD(INVALID, NONE, 0, 0, "Reserved", false, 0),
> > > +};
> > > +
> > > static int cxl_mem_wait_for_doorbell(struct cxl_mem *cxlm)
> > > {
> > > const int timeout = msecs_to_jiffies(2000);
> > > @@ -268,8 +312,53 @@ static int cxl_mem_open(struct inode *inode, struct file *file)
> > > return 0;
> > > }
> > >
> > > +static int cxl_mem_count_commands(void)
> > > +{
> > > + int i, n = 0;
> > > +
> > > + for (i = 0; i < ARRAY_SIZE(mem_commands); i++) {
> > > + struct cxl_mem_command *c = &mem_commands[i];
> > > +
> > > + if (c->enable)
> > > + n++;
> > > + }
> > > +
> > > + return n;
> > > +}
> > > +
> > > static long cxl_mem_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
> > > {
> > > + if (cmd == CXL_MEM_QUERY_COMMANDS) {
> > > + struct cxl_mem_query_commands __user *q = (void __user *)arg;
> > > + u32 n_commands;
> > > + int i, j;
> > > +
> > > + if (get_user(n_commands, (u32 __user *)arg))
> > > + return -EFAULT;
> > > +
> > > + if (n_commands == 0)
> > > + return put_user(cxl_mem_count_commands(),
> > > + (u32 __user *)arg);
> > > +
> > > + for (i = 0, j = 0;
> > > + i < ARRAY_SIZE(mem_commands) && j < n_commands; i++) {
> > > + struct cxl_mem_command *c = &mem_commands[i];
> > > + const struct cxl_command_info *info = &c->info;
> > > +
> > > + if (!c->enable)
> > > + continue;
> > > +
> > > + if (copy_to_user(&q->commands[j], info, sizeof(*info)))
> > > + return -EFAULT;
> > > +
> > > + if (copy_to_user(&q->commands[j].name, info->name,
> > > + strlen(info->name)))
> > > + return -EFAULT;
> >
> > Not sure why this is needed, see comment below about @name in
> > cxl_mem_query_commands.
> >
> > > +
> > > + j++;
> > > + }
> > > + }
> > > +
> > > return -ENOTTY;
> > > }
> > >
> > > diff --git a/include/uapi/linux/cxl_mem.h b/include/uapi/linux/cxl_mem.h
> > > new file mode 100644
> > > index 000000000000..1d1e143f98ec
> > > --- /dev/null
> > > +++ b/include/uapi/linux/cxl_mem.h
> > > @@ -0,0 +1,102 @@
> > > +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
> > > +/*
> > > + * CXL IOCTLs for Memory Devices
> > > + */
> > > +
> > > +#ifndef _UAPI_CXL_MEM_H_
> > > +#define _UAPI_CXL_MEM_H_
> > > +
> > > +#if defined(__cplusplus)
> > > +extern "C" {
> > > +#endif
> > > +
> > > +/**
> > > + * DOC: UAPI
> > > + *
> > > + * CXL memory devices expose UAPI to have a standard user interface.
> > > + * Userspace can refer to these structure definitions and UAPI formats
> > > + * to communicate to driver
> > > + */
> > > +
> > > +#define CXL_MEM_QUERY_COMMANDS _IOR('C', 1, struct cxl_mem_query_commands)
> > > +
> > > +#define CXL_MEM_COMMAND_NAME_LENGTH 32
> > > +
> > > +/**
> > > + * struct cxl_command_info - Command information returned from a query.
> > > + * @id: ID number for the command.
> > > + * @flags: Flags that specify command behavior.
> > > + *
> > > + * - CXL_MEM_COMMAND_FLAG_TAINT: Using this command will taint the kernel.
> > > + * @size_in: Expected input size, or -1 if variable length.
> > > + * @size_out: Expected output size, or -1 if variable length.
> > > + * @name: Name describing the command.
> > > + *
> > > + * Represents a single command that is supported by both the driver and the
> > > + * hardware. The is returned as part of an array from the query ioctl. The
> > > + * following would be a command named "foobar" that takes a variable length
> > > + * input and returns 0 bytes of output.
> > > + *
> > > + * - @id = 10
> > > + * - @name = foobar
> > > + * - @flags = 0
> > > + * - @size_in = -1
> > > + * - @size_out = 0
> > > + *
> > > + * See struct cxl_mem_query_commands.
> > > + */
> > > +struct cxl_command_info {
> > > + __u32 id;
> > > +#define CXL_MEM_COMMAND_ID_INVALID 0
> > > +
> > > + __u32 flags;
> > > +#define CXL_MEM_COMMAND_FLAG_NONE 0
> > > +#define CXL_MEM_COMMAND_FLAG_TAINT BIT(0)
> > > +
> > > + __s32 size_in;
> > > + __s32 size_out;
> > > +
> > > + char name[32];
> >
> > Why does the name for a command need to be shuffled back and forth
> > over the ioctl interface. Can't this be handled by a static lookup
> > table defined in the header?
> >
>
> I was thinking of cases where the userspace application doesn't match the
> current kernel's UAPI and giving the driver flexibility to return whatever.

How / why would the application by looking at @name for UAPI compatibility?

> OTTOMH, I also can't think of a way to do this if you want to do define the
> table sparsely though. Do you have ideas for that?

I don't think the name lookup would be sparse. i.e. it would be ok for
mem_commands to not have an entry for everything in the name lookup
table. As for defining the table it could use C preprocessor trick
popularized by Steven Rostedt:

#define CMDS \
C(CMD1, "command one"), \
C(CMD2, "command two") \
#undef C
#define C(a, b) a
enum commands_enum { CMDS };
#undef C
#define C(a, b) { b }
static struct {
const char *name;
} commands[] = { CMDS };
#undef C

...then there's no way for the command ids to get out of sync with the names.