Re: [PATCH v6 3/5] fw_cfg: do DMA read operation

From: Michael S. Tsirkin
Date: Thu Nov 16 2017 - 11:10:06 EST


On Thu, Nov 16, 2017 at 08:17:12AM -0500, Marc-André Lureau wrote:
> Hi
>
> ----- Original Message -----
> > On Mon, Nov 13, 2017 at 08:29:56PM +0100, Marc-André Lureau wrote:
> > > Modify fw_cfg_read_blob() to use DMA if the device supports it.
> > > Return errors, because the operation may fail.
> > >
> > > To avoid polling with unbound amount of time, the DMA operation is
> > > expected to complete within 200ms, or will return ETIME error.
> > >
> > > We may want to switch all the *buf addresses to use only kmalloc'ed
> > > buffers (instead of using stack/image addresses with dma=false).
> > >
> > > Signed-off-by: Marc-André Lureau <marcandre.lureau@xxxxxxxxxx>
> > > Reviewed-by: Gabriel Somlo <somlo@xxxxxxx>
> > > ---
> > > drivers/firmware/qemu_fw_cfg.c | 154
> > > ++++++++++++++++++++++++++++++++++++-----
> > > 1 file changed, 137 insertions(+), 17 deletions(-)
> > >
> > > diff --git a/drivers/firmware/qemu_fw_cfg.c
> > > b/drivers/firmware/qemu_fw_cfg.c
> > > index 1f3e8545dab7..2ac4cd869fe6 100644
> > > --- a/drivers/firmware/qemu_fw_cfg.c
> > > +++ b/drivers/firmware/qemu_fw_cfg.c
> > > @@ -33,6 +33,8 @@
> > > #include <linux/slab.h>
> > > #include <linux/io.h>
> > > #include <linux/ioport.h>
> > > +#include <linux/delay.h>
> > > +#include <linux/dma-mapping.h>
> > >
> > > MODULE_AUTHOR("Gabriel L. Somlo <somlo@xxxxxxx>");
> > > MODULE_DESCRIPTION("QEMU fw_cfg sysfs support");
> > > @@ -43,12 +45,26 @@ MODULE_LICENSE("GPL");
> > > #define FW_CFG_ID 0x01
> > > #define FW_CFG_FILE_DIR 0x19
> > >
> > > +#define FW_CFG_VERSION_DMA 0x02
> > > +#define FW_CFG_DMA_CTL_ERROR 0x01
> > > +#define FW_CFG_DMA_CTL_READ 0x02
> > > +#define FW_CFG_DMA_CTL_SKIP 0x04
> > > +#define FW_CFG_DMA_CTL_SELECT 0x08
> > > +#define FW_CFG_DMA_CTL_WRITE 0x10
> > > +#define FW_CFG_DMA_TIMEOUT 200 /* ms */
> > > +
> > > /* size in bytes of fw_cfg signature */
> > > #define FW_CFG_SIG_SIZE 4
> > >
> > > /* fw_cfg "file name" is up to 56 characters (including terminating nul)
> > > */
> > > #define FW_CFG_MAX_FILE_PATH 56
> > >
> > > +/* platform device for dma mapping */
> > > +static struct device *dev;
> > > +
> > > +/* fw_cfg revision attribute, in /sys/firmware/qemu_fw_cfg top-level dir.
> > > */
> > > +static u32 fw_cfg_rev;
> > > +
> > > /* fw_cfg file directory entry type */
> > > struct fw_cfg_file {
> > > u32 size;
> > > @@ -57,6 +73,12 @@ struct fw_cfg_file {
> > > char name[FW_CFG_MAX_FILE_PATH];
> > > };
> > >
> > > +struct fw_cfg_dma {
> > > + u32 control;
> > > + u32 length;
> > > + u64 address;
> > > +} __packed;
> > > +
> > > /* fw_cfg device i/o register addresses */
> > > static bool fw_cfg_is_mmio;
> > > static phys_addr_t fw_cfg_p_base;
> > > @@ -75,12 +97,93 @@ static inline u16 fw_cfg_sel_endianness(u16 key)
> > > return fw_cfg_is_mmio ? cpu_to_be16(key) : cpu_to_le16(key);
> > > }
> > >
> > > +static inline bool fw_cfg_dma_enabled(void)
> > > +{
> > > + return fw_cfg_rev & FW_CFG_VERSION_DMA && fw_cfg_reg_dma;
> > > +}
> > > +
> > > +static bool fw_cfg_wait_for_control(struct fw_cfg_dma *d, unsigned long
> > > timeout)
> > > +{
> > > + ktime_t start;
> > > + ktime_t stop;
> > > +
> > > + start = ktime_get();
> > > + stop = ktime_add(start, ms_to_ktime(timeout));
> > > +
> > > + do {
> > > + if ((be32_to_cpu(d->control) & ~FW_CFG_DMA_CTL_ERROR) == 0)
> > > + return true;
> > > +
> > > + usleep_range(50, 100);
> >
> > BTW it's not nice that this is uninterruptible. I think we need a
> > variant of usleep_range that is interruptible and call that here.
> > Thoughts?
> >
>
> This usleep_range() pattern is pretty common apparently.

In kernel - right, but doing it from userspace context means you can't
kill userspace until timeout expires.

>
> (and it's probably better than calling yield() in a loop, like v1-3 did)
> >
> > > + } while (ktime_before(ktime_get(), stop));
> > > +
> > > + return false;
> > > +}
> > > +
> > > +static ssize_t fw_cfg_dma_transfer(void *address, u32 length, u32 control)
> > > +{
> > > + dma_addr_t dma_addr = 0;
> > > + struct fw_cfg_dma *d;
> > > + dma_addr_t dma;
> > > + ssize_t ret = length;
> > > + enum dma_data_direction dir =
> > > + (control & FW_CFG_DMA_CTL_READ ? DMA_FROM_DEVICE : 0);
> > > +
> > > + if (address && length) {
> > > + dma_addr = dma_map_single(dev, address, length, dir);
> > > + if (dma_mapping_error(NULL, dma_addr)) {
> > > + WARN(1, "%s: failed to map address\n", __func__);
> > > + return -EFAULT;
> > > + }
> > > + }
> > > +
> > > + d = kmalloc(sizeof(*d), GFP_KERNEL | GFP_DMA);
> > > + if (!d) {
> > > + ret = -ENOMEM;
> > > + goto end;
> > > + }
> > > +
> > > + dma = dma_map_single(dev, d, sizeof(*d), DMA_BIDIRECTIONAL);
> > > + if (dma_mapping_error(NULL, dma)) {
> > > + WARN(1, "%s: failed to map fw_cfg_dma\n", __func__);
> > > + ret = -EFAULT;
> > > + goto end;
> > > + }
> > > +
> > > + *d = (struct fw_cfg_dma) {
> > > + .address = cpu_to_be64(dma_addr),
> > > + .length = cpu_to_be32(length),
> > > + .control = cpu_to_be32(control)
> > > + };
> > > +
> > > + iowrite32be((u64)dma >> 32, fw_cfg_reg_dma);
> > > + iowrite32be(dma, fw_cfg_reg_dma + 4);
> > > +
> > > + if (!fw_cfg_wait_for_control(d, FW_CFG_DMA_TIMEOUT)) {
> > > + WARN(1, "%s: timeout", __func__);
> > > + ret = -ETIME;
> > > + } else if (be32_to_cpu(d->control) & FW_CFG_DMA_CTL_ERROR) {
> > > + ret = -EIO;
> > > + }
> > > +
> > > + dma_unmap_single(dev, dma, sizeof(*d), DMA_BIDIRECTIONAL);
> > > +
> > > +end:
> > > + kfree(d);
> > > + if (dma_addr)
> > > + dma_unmap_single(dev, dma_addr, length, dir);
> >
> > So if host was delayed what this does is corrupt guest memory.
>
> By "delayed" you mean fw_cfg_wait_for_control() timed out?
>
> Hmm, perhaps we should loop infinitely? that's also what the firmware does.
>
> By the way, I failed to understand vcpu guest & qemu interaction here, shouldn't the vcpu thread be running the qemu dma_transfer until completion before returning to guest?
>
> > IMHO things are already bad, let's at least keep it mapped
> > and allocated. Also what will heppen on next access attempt?
>
> > Maybe we need to prevent further attempts.
> > And maybe fw cfg needs a reset register so we can recover
> > faster.
>
> Or perhaps since it's all software, it should never fail? ;) That's apparently the device design.

As far as I can see, that is true. IOW the time-out is not needed,
if it's not completed on the first attempt, you can fail
immediately.


But a bigger issue is it seems to trigger failures for people.

> >
> >
> > > +
> > > + return ret;
> > > +}
> > > +
> > > /* read chunk of given fw_cfg blob (caller responsible for sanity-check)
> > > */
> > > -static inline void fw_cfg_read_blob(u16 key,
> > > - void *buf, loff_t pos, size_t count)
> > > +static ssize_t fw_cfg_read_blob(u16 key,
> > > + void *buf, loff_t pos, size_t count,
> > > + bool dma)
> > > {
> > > u32 glk = -1U;
> > > acpi_status status;
> > > + ssize_t ret = count;
> > >
> > > /* If we have ACPI, ensure mutual exclusion against any potential
> > > * device access by the firmware, e.g. via AML methods:
> > > @@ -90,17 +193,36 @@ static inline void fw_cfg_read_blob(u16 key,
> > > /* Should never get here */
> > > WARN(1, "fw_cfg_read_blob: Failed to lock ACPI!\n");
> > > memset(buf, 0, count);
> > > - return;
> > > + return -EINVAL;
> > > }
> > >
> > > mutex_lock(&fw_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);
> > > + if (dma && fw_cfg_dma_enabled()) {
> > > + if (pos == 0) {
> > > + ret = fw_cfg_dma_transfer(buf, count, key << 16
> > > + | FW_CFG_DMA_CTL_SELECT
> > > + | FW_CFG_DMA_CTL_READ);
> > > + } else {
> > > + iowrite16(fw_cfg_sel_endianness(key), fw_cfg_reg_ctrl);
> > > + ret = fw_cfg_dma_transfer(NULL, pos, FW_CFG_DMA_CTL_SKIP);
> > > + if (ret < 0)
> > > + goto end;
> > > + ret = fw_cfg_dma_transfer(buf, count,
> > > + FW_CFG_DMA_CTL_READ);
> > > + }
> > > + } else {
> > > + 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);
> > > + }
> > > +
> > > +end:
> > > mutex_unlock(&fw_cfg_dev_lock);
> > >
> > > acpi_release_global_lock(glk);
> > > +
> > > + return ret;
> > > }
> > >
> > > /* clean up fw_cfg device i/o */
> > > @@ -192,7 +314,7 @@ static int fw_cfg_do_platform_probe(struct
> > > platform_device *pdev)
> > > #endif
> > >
> > > /* verify fw_cfg device signature */
> > > - fw_cfg_read_blob(FW_CFG_SIGNATURE, sig, 0, FW_CFG_SIG_SIZE);
> > > + fw_cfg_read_blob(FW_CFG_SIGNATURE, sig, 0, FW_CFG_SIG_SIZE, false);
> > > if (memcmp(sig, "QEMU", FW_CFG_SIG_SIZE) != 0) {
> > > fw_cfg_io_cleanup();
> > > return -ENODEV;
> > > @@ -201,9 +323,6 @@ static int fw_cfg_do_platform_probe(struct
> > > platform_device *pdev)
> > > return 0;
> > > }
> > >
> > > -/* fw_cfg revision attribute, in /sys/firmware/qemu_fw_cfg top-level dir.
> > > */
> > > -static u32 fw_cfg_rev;
> > > -
> > > static ssize_t fw_cfg_showrev(struct kobject *k, struct attribute *a, char
> > > *buf)
> > > {
> > > return sprintf(buf, "%u\n", fw_cfg_rev);
> > > @@ -351,8 +470,7 @@ static ssize_t fw_cfg_sysfs_read_raw(struct file *filp,
> > > struct kobject *kobj,
> > > if (count > entry->f.size - pos)
> > > count = entry->f.size - pos;
> > >
> > > - fw_cfg_read_blob(entry->f.select, buf, pos, count);
> > > - return count;
> > > + return fw_cfg_read_blob(entry->f.select, buf, pos, count, true);
> > > }
> > >
> > > static struct bin_attribute fw_cfg_sysfs_attr_raw = {
> > > @@ -505,7 +623,7 @@ static int fw_cfg_register_dir_entries(void)
> > > struct fw_cfg_file *dir;
> > > size_t dir_size;
> > >
> > > - fw_cfg_read_blob(FW_CFG_FILE_DIR, &count, 0, sizeof(count));
> > > + fw_cfg_read_blob(FW_CFG_FILE_DIR, &count, 0, sizeof(count), false);
> > > count = be32_to_cpu(count);
> > > dir_size = count * sizeof(struct fw_cfg_file);
> > >
> > > @@ -513,7 +631,7 @@ static int fw_cfg_register_dir_entries(void)
> > > if (!dir)
> > > return -ENOMEM;
> > >
> > > - fw_cfg_read_blob(FW_CFG_FILE_DIR, dir, sizeof(count), dir_size);
> > > + fw_cfg_read_blob(FW_CFG_FILE_DIR, dir, sizeof(count), dir_size, true);
> > >
> > > for (i = 0; i < count; i++) {
> > > dir[i].size = be32_to_cpu(dir[i].size);
> > > @@ -544,9 +662,10 @@ static int fw_cfg_sysfs_probe(struct platform_device
> > > *pdev)
> > > * one fw_cfg device exist system-wide, so if one was already found
> > > * earlier, we might as well stop here.
> > > */
> > > - if (fw_cfg_sel_ko)
> > > + if (dev)
> > > return -EBUSY;
> > >
> > > + dev = &pdev->dev;
> > > /* create by_key and by_name subdirs of /sys/firmware/qemu_fw_cfg/ */
> > > err = -ENOMEM;
> > > fw_cfg_sel_ko = kobject_create_and_add("by_key", fw_cfg_top_ko);
> > > @@ -562,7 +681,7 @@ static int fw_cfg_sysfs_probe(struct platform_device
> > > *pdev)
> > > goto err_probe;
> > >
> > > /* get revision number, add matching top-level attribute */
> > > - fw_cfg_read_blob(FW_CFG_ID, &fw_cfg_rev, 0, sizeof(fw_cfg_rev));
> > > + fw_cfg_read_blob(FW_CFG_ID, &fw_cfg_rev, 0, sizeof(fw_cfg_rev), false);
> > > fw_cfg_rev = le32_to_cpu(fw_cfg_rev);
> > > err = sysfs_create_file(fw_cfg_top_ko, &fw_cfg_rev_attr.attr);
> > > if (err)
> > > @@ -587,6 +706,7 @@ static int fw_cfg_sysfs_probe(struct platform_device
> > > *pdev)
> > > err_name:
> > > fw_cfg_kobj_cleanup(fw_cfg_sel_ko);
> > > err_sel:
> > > + dev = NULL;
> > > return err;
> > > }
> > >
> > > --
> > > 2.15.0.125.g8f49766d64
> >