Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Tue, 21 Nov 2017 06:17:46 -0500 (EST)
From:   =?utf-8?Q?Marc-Andr=C3=A9?= Lureau <marcandre.lureau@redhat.com>
To:     "Michael S. Tsirkin" <mst@redhat.com>
Cc:     linux-kernel@vger.kernel.org, qemu-devel@nongnu.org, somlo@cmu.edu,
        xiaolong ye <xiaolong.ye@intel.com>
Message-ID: <1367386885.43304805.1511263066456.JavaMail.zimbra@redhat.com>
In-Reply-To: <20171120232407-mutt-send-email-mst@kernel.org>
References: <20171120095519.15214-1-marcandre.lureau@redhat.com> <20171120095519.15214-4-marcandre.lureau@redhat.com> <20171120232407-mutt-send-email-mst@kernel.org>
Subject: Re: [PATCH v7 3/5] fw_cfg: do DMA read operation
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8BIT
Thread-Topic: fw_cfg: do DMA read operation
Thread-Index: iBmHRTsCpSUHIQfuBP4DW3k38J7nCw==
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi

----- Original Message -----
> On Mon, Nov 20, 2017 at 10:55:17AM +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.
> > 
> > The DMA operation is expected to run synchronously with today qemu,
> > but the specification states that it may become async, so we run
> > "control" field check in a loop for eventual changes.
> > 
> > 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@redhat.com>
> > ---
> >  drivers/firmware/qemu_fw_cfg.c | 140
> >  ++++++++++++++++++++++++++++++++++++-----
> >  1 file changed, 123 insertions(+), 17 deletions(-)
> > 
> > diff --git a/drivers/firmware/qemu_fw_cfg.c
> > b/drivers/firmware/qemu_fw_cfg.c
> > index 740df0df2260..e0fe6ff037c3 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@cmu.edu>");
> >  MODULE_DESCRIPTION("QEMU fw_cfg sysfs support");
> > @@ -43,12 +45,25 @@ 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
> > +
> >  /* 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 +72,12 @@ 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;
> 
> Drop __packed please. It causes many gcc versions to do insane things.
> Can be a patch on top.

Oh? I think __packed should translate to __attribute__((packed)) (include/linux/compiler-gcc.h), there would be serious problems if gcc "do insane things" with it.

> 
> 
> > @@ -75,12 +96,79 @@ 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;
> > +}
> > +
> > +/* qemu fw_cfg device is sync today, but spec says it may become async */
> > +static void fw_cfg_wait_for_control(struct fw_cfg_dma *d, dma_addr_t dma)
> > +{
> > +	do {
> > +		dma_sync_single_for_cpu(dev, dma, sizeof(*d), DMA_FROM_DEVICE);
> > +		if ((be32_to_cpu(d->control) & ~FW_CFG_DMA_CTL_ERROR) == 0)
> > +			return;
> > +
> > +		usleep_range(50, 100);
> 
> And since in practice we never get to this line,
> maybe we should just go back to yield here.

Or cond_resched() ?

> 
> > +	} while (true);
> > +}
> > +
> > +static ssize_t fw_cfg_dma_transfer(void *address, u32 length, u32 control)
> > +{
> > +	dma_addr_t dma_addr = 0;
> > +	static 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 = (struct fw_cfg_dma) {
> > +		.address = cpu_to_be64(dma_addr),
> > +		.length = cpu_to_be32(length),
> > +		.control = cpu_to_be32(control)
> > +	};
> > +
> > +	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;
> > +	}
> > +
> > +	iowrite32be((u64)dma >> 32, fw_cfg_reg_dma);
> > +	iowrite32be(dma, fw_cfg_reg_dma + 4);
> > +
> > +	fw_cfg_wait_for_control(&d, dma);
> > +
> > +	if (be32_to_cpu(d.control) & FW_CFG_DMA_CTL_ERROR) {
> > +		ret = -EIO;
> > +	}
> > +
> > +	dma_unmap_single(dev, dma, sizeof(d), DMA_BIDIRECTIONAL);
> > +
> > +end:
> > +	if (dma_addr)
> > +		dma_unmap_single(dev, dma_addr, length, dir);
> > +
> > +	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 +178,36 @@ 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 +299,7 @@ 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 +308,6 @@ 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 +455,7 @@ 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 +608,7 @@ 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 +616,7 @@ 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 +647,10 @@ 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 +666,7 @@ 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 +691,7 @@ 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;
> >  }
> >  
> > @@ -598,6 +703,7 @@ static int fw_cfg_sysfs_remove(struct platform_device
> > *pdev)
> >  	fw_cfg_io_cleanup();
> >  	fw_cfg_kset_unregister_recursive(fw_cfg_fname_kset);
> >  	fw_cfg_kobj_cleanup(fw_cfg_sel_ko);
> > +	dev = NULL;
> >  	return 0;
> >  }
> >  
> > --
> > 2.15.0.277.ga3d2ad2c43
> 

From 1584623394649336894@xxx Mon Nov 20 21:50:32 +0000 2017
X-GM-THRID: 1584578502964492301
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread