linux/drivers/usb/usbip/stub_tx.c

454 lines
11 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0+
/*
* Copyright (C) 2003-2008 Takahiro Hirofuchi
*/
#include <linux/kthread.h>
#include <linux/socket.h>
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
#include <linux/scatterlist.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include "usbip_common.h"
#include "stub.h"
/* be in spin_lock_irqsave(&sdev->priv_lock, flags) */
void stub_enqueue_ret_unlink(struct stub_device *sdev, __u32 seqnum,
__u32 status)
{
struct stub_unlink *unlink;
unlink = kzalloc(sizeof(struct stub_unlink), GFP_ATOMIC);
if (!unlink) {
usbip_event_add(&sdev->ud, VDEV_EVENT_ERROR_MALLOC);
return;
}
unlink->seqnum = seqnum;
unlink->status = status;
list_add_tail(&unlink->list, &sdev->unlink_tx);
}
/**
* stub_complete - completion handler of a usbip urb
* @urb: pointer to the urb completed
*
* When a urb has completed, the USB core driver calls this function mostly in
* the interrupt context. To return the result of a urb, the completed urb is
* linked to the pending list of returning.
*
*/
void stub_complete(struct urb *urb)
{
struct stub_priv *priv = (struct stub_priv *) urb->context;
struct stub_device *sdev = priv->sdev;
unsigned long flags;
usbip_dbg_stub_tx("complete! status %d\n", urb->status);
switch (urb->status) {
case 0:
/* OK */
break;
case -ENOENT:
dev_info(&urb->dev->dev,
"stopped by a call to usb_kill_urb() because of cleaning up a virtual connection\n");
return;
case -ECONNRESET:
dev_info(&urb->dev->dev,
"unlinked by a call to usb_unlink_urb()\n");
break;
case -EPIPE:
dev_info(&urb->dev->dev, "endpoint %d is stalled\n",
usb_pipeendpoint(urb->pipe));
break;
case -ESHUTDOWN:
dev_info(&urb->dev->dev, "device removed?\n");
break;
default:
dev_info(&urb->dev->dev,
"urb completion with non-zero status %d\n",
urb->status);
break;
}
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
/*
* If the server breaks single SG request into the several URBs, the
* URBs must be reassembled before sending completed URB to the vhci.
* Don't wake up the tx thread until all the URBs are completed.
*/
if (priv->sgl) {
priv->completed_urbs++;
/* Only save the first error status */
if (urb->status && !priv->urb_status)
priv->urb_status = urb->status;
if (priv->completed_urbs < priv->num_urbs)
return;
}
/* link a urb to the queue of tx. */
spin_lock_irqsave(&sdev->priv_lock, flags);
usbip: safe completion against unbind operation This patch adds a code fragment to ignore completing URBs in closing connection. Regarding this patch, 2 execution contexts are related. 1) stub_tx.c: stub_complete() which is called from USB core 1-1) add to unlink list and free URB or 1-2) move to tx list 2) stub_dev.c: stub_shutdown_connection() which is invoked by unbind operation through sysfs. 2-1) stop TX/RX threads 2-2) close TCP connection and set ud.tcp_socket to NULL 2-3) cleanup pending URBs by stub_device_cleanup_urbs(sdev) 2-4) free unlink list (no lock) In the race condition, URBs which will be cleared in 2-3) may be handled in 1). In case 1-1), it will not be transferred bcause tx threads are stooped in 2-1). In case 1-2), may be freed in 2-4). With this patch, after 2-2), completing URBs in 1) will not be handled and cleared in 2-3). The kernel log with this patch is as below. kernel: usbip_core: usbip_kernel_unlink:792: shutting down tcp_socket ef61d980 kernel: usbip-host 1-3: free sdev f5df6180 kernel: usbip-host 1-3: free urb f5df6700 kernel: usbip-host 1-3: Enter kernel: usbip_core: usbip_stop_eh:132: usbip_eh waiting completion 5 kernel: usbip_host: stub_complete:71: complete! status 0 kernel: usbip_host: stub_complete:102: ignore urb for closed connection e725fc00 (*) kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: free urb e725fc00 (**) kernel: usbip-host 1-3: free urb e725e000 kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: free urb e725f800 kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: free urb e725e800 kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: device reset kernel: usbip-host 1-3: lock for reset kernel: usbip_host: store_match_busid:178: del busid 1-3 kernel: uvcvideo: Found UVC 1.00 device Venus USB2.0 Camera (056e:700a) kernel: input: Venus USB2.0 Camera as /devices/pci0000:00/0000:00:1a.7/usb1/1-3/1-3:1.0/input/input22 (*) skipped with this patch in completion (**) released in 2-3 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-04-27 14:35:53 +08:00
if (sdev->ud.tcp_socket == NULL) {
usbip_dbg_stub_tx("ignore urb for closed connection\n");
usbip: safe completion against unbind operation This patch adds a code fragment to ignore completing URBs in closing connection. Regarding this patch, 2 execution contexts are related. 1) stub_tx.c: stub_complete() which is called from USB core 1-1) add to unlink list and free URB or 1-2) move to tx list 2) stub_dev.c: stub_shutdown_connection() which is invoked by unbind operation through sysfs. 2-1) stop TX/RX threads 2-2) close TCP connection and set ud.tcp_socket to NULL 2-3) cleanup pending URBs by stub_device_cleanup_urbs(sdev) 2-4) free unlink list (no lock) In the race condition, URBs which will be cleared in 2-3) may be handled in 1). In case 1-1), it will not be transferred bcause tx threads are stooped in 2-1). In case 1-2), may be freed in 2-4). With this patch, after 2-2), completing URBs in 1) will not be handled and cleared in 2-3). The kernel log with this patch is as below. kernel: usbip_core: usbip_kernel_unlink:792: shutting down tcp_socket ef61d980 kernel: usbip-host 1-3: free sdev f5df6180 kernel: usbip-host 1-3: free urb f5df6700 kernel: usbip-host 1-3: Enter kernel: usbip_core: usbip_stop_eh:132: usbip_eh waiting completion 5 kernel: usbip_host: stub_complete:71: complete! status 0 kernel: usbip_host: stub_complete:102: ignore urb for closed connection e725fc00 (*) kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: free urb e725fc00 (**) kernel: usbip-host 1-3: free urb e725e000 kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: free urb e725f800 kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: free urb e725e800 kernel: usbip_host: stub_complete:71: complete! status -2 kernel: usbip-host 1-3: stopped by a call to usb_kill_urb() because of cleaning up a virtual connection kernel: usbip-host 1-3: device reset kernel: usbip-host 1-3: lock for reset kernel: usbip_host: store_match_busid:178: del busid 1-3 kernel: uvcvideo: Found UVC 1.00 device Venus USB2.0 Camera (056e:700a) kernel: input: Venus USB2.0 Camera as /devices/pci0000:00/0000:00:1a.7/usb1/1-3/1-3:1.0/input/input22 (*) skipped with this patch in completion (**) released in 2-3 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-04-27 14:35:53 +08:00
/* It will be freed in stub_device_cleanup_urbs(). */
} else if (priv->unlinking) {
stub_enqueue_ret_unlink(sdev, priv->seqnum, urb->status);
stub_free_priv_and_urb(priv);
} else {
list_move_tail(&priv->list, &sdev->priv_tx);
}
spin_unlock_irqrestore(&sdev->priv_lock, flags);
/* wake up tx_thread */
wake_up(&sdev->tx_waitq);
}
static inline void setup_base_pdu(struct usbip_header_basic *base,
__u32 command, __u32 seqnum)
{
base->command = command;
base->seqnum = seqnum;
base->devid = 0;
base->ep = 0;
base->direction = 0;
}
static void setup_ret_submit_pdu(struct usbip_header *rpdu, struct urb *urb)
{
struct stub_priv *priv = (struct stub_priv *) urb->context;
setup_base_pdu(&rpdu->base, USBIP_RET_SUBMIT, priv->seqnum);
usbip_pack_pdu(rpdu, urb, USBIP_RET_SUBMIT, 1);
}
static void setup_ret_unlink_pdu(struct usbip_header *rpdu,
struct stub_unlink *unlink)
{
setup_base_pdu(&rpdu->base, USBIP_RET_UNLINK, unlink->seqnum);
rpdu->u.ret_unlink.status = unlink->status;
}
static struct stub_priv *dequeue_from_priv_tx(struct stub_device *sdev)
{
unsigned long flags;
struct stub_priv *priv, *tmp;
spin_lock_irqsave(&sdev->priv_lock, flags);
list_for_each_entry_safe(priv, tmp, &sdev->priv_tx, list) {
list_move_tail(&priv->list, &sdev->priv_free);
spin_unlock_irqrestore(&sdev->priv_lock, flags);
return priv;
}
spin_unlock_irqrestore(&sdev->priv_lock, flags);
return NULL;
}
static int stub_send_ret_submit(struct stub_device *sdev)
{
unsigned long flags;
struct stub_priv *priv, *tmp;
struct msghdr msg;
size_t txsize;
size_t total_size = 0;
while ((priv = dequeue_from_priv_tx(sdev)) != NULL) {
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
struct urb *urb = priv->urbs[0];
struct usbip_header pdu_header;
struct usbip_iso_packet_descriptor *iso_buffer = NULL;
struct kvec *iov = NULL;
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
struct scatterlist *sg;
u32 actual_length = 0;
int iovnum = 0;
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
int ret;
int i;
txsize = 0;
memset(&pdu_header, 0, sizeof(pdu_header));
memset(&msg, 0, sizeof(msg));
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
if (urb->actual_length > 0 && !urb->transfer_buffer &&
!urb->num_sgs) {
dev_err(&sdev->udev->dev,
"urb: actual_length %d transfer_buffer null\n",
urb->actual_length);
return -1;
}
if (usb_pipetype(urb->pipe) == PIPE_ISOCHRONOUS)
iovnum = 2 + urb->number_of_packets;
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
else if (usb_pipein(urb->pipe) && urb->actual_length > 0 &&
urb->num_sgs)
iovnum = 1 + urb->num_sgs;
else if (usb_pipein(urb->pipe) && priv->sgl)
iovnum = 1 + priv->num_urbs;
else
iovnum = 2;
iov = kcalloc(iovnum, sizeof(struct kvec), GFP_KERNEL);
if (!iov) {
usbip_event_add(&sdev->ud, SDEV_EVENT_ERROR_MALLOC);
return -1;
}
iovnum = 0;
/* 1. setup usbip_header */
setup_ret_submit_pdu(&pdu_header, urb);
usbip_dbg_stub_tx("setup txdata seqnum: %d\n",
pdu_header.base.seqnum);
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
if (priv->sgl) {
for (i = 0; i < priv->num_urbs; i++)
actual_length += priv->urbs[i]->actual_length;
pdu_header.u.ret_submit.status = priv->urb_status;
pdu_header.u.ret_submit.actual_length = actual_length;
}
usbip_header_correct_endian(&pdu_header, 1);
iov[iovnum].iov_base = &pdu_header;
iov[iovnum].iov_len = sizeof(pdu_header);
iovnum++;
txsize += sizeof(pdu_header);
/* 2. setup transfer buffer */
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
if (usb_pipein(urb->pipe) && priv->sgl) {
/* If the server split a single SG request into several
* URBs because the server's HCD doesn't support SG,
* reassemble the split URB buffers into a single
* return command.
*/
for (i = 0; i < priv->num_urbs; i++) {
iov[iovnum].iov_base =
priv->urbs[i]->transfer_buffer;
iov[iovnum].iov_len =
priv->urbs[i]->actual_length;
iovnum++;
}
txsize += actual_length;
} else if (usb_pipein(urb->pipe) &&
usb_pipetype(urb->pipe) != PIPE_ISOCHRONOUS &&
urb->actual_length > 0) {
usbip: Implement SG support to vhci-hcd and stub driver There are bugs on vhci with usb 3.0 storage device. In USB, each SG list entry buffer should be divisible by the bulk max packet size. But with native SG support, this problem doesn't matter because the SG buffer is treated as contiguous buffer. But without native SG support, USB storage driver breaks SG list into several URBs and the error occurs because of a buffer size of URB that cannot be divided by the bulk max packet size. The error situation is as follows. When USB Storage driver requests 31.5 KB data and has SG list which has 3584 bytes buffer followed by 7 4096 bytes buffer for some reason. USB Storage driver splits this SG list into several URBs because VHCI doesn't support SG and sends them separately. So the first URB buffer size is 3584 bytes. When receiving data from device, USB 3.0 device sends data packet of 1024 bytes size because the max packet size of BULK pipe is 1024 bytes. So device sends 4096 bytes. But the first URB buffer has only 3584 bytes buffer size. So host controller terminates the transfer even though there is more data to receive. So, vhci needs to support SG transfer to prevent this error. In this patch, vhci supports SG regardless of whether the server's host controller supports SG or not, because stub driver splits SG list into several URBs if the server's host controller doesn't support SG. To support SG, vhci sets URB_DMA_MAP_SG flag in urb->transfer_flags if URB has SG list and this flag will tell stub driver to use SG list. After receiving urb from stub driver, vhci clear URB_DMA_MAP_SG flag to avoid unnecessary DMA unmapping in HCD. vhci sends each SG list entry to stub driver. Then, stub driver sees the total length of the buffer and allocates SG table and pages according to the total buffer length calling sgl_alloc(). After stub driver receives completed URB, it again sends each SG list entry to vhci. If the server's host controller doesn't support SG, stub driver breaks a single SG request into several URBs and submits them to the server's host controller. When all the split URBs are completed, stub driver reassembles the URBs into a single return command and sends it to vhci. Moreover, in the situation where vhci supports SG, but stub driver does not, or vice versa, usbip works normally. Because there is no protocol modification, there is no problem in communication between server and client even if the one has a kernel without SG support. In the case of vhci supports SG and stub driver doesn't, because vhci sends only the total length of the buffer to stub driver as it did before the patch applied, stub driver only needs to allocate the required length of buffers using only kmalloc() regardless of whether vhci supports SG or not. But stub driver has to allocate buffer with kmalloc() as much as the total length of SG buffer which is quite huge when vhci sends SG request, so it has overhead in buffer allocation in this situation. If stub driver needs to send data buffer to vhci because of IN pipe, stub driver also sends only total length of buffer as metadata and then sends real data as vhci does. Then vhci receive data from stub driver and store it to the corresponding buffer of SG list entry. And for the case of stub driver supports SG and vhci doesn't, since the USB storage driver checks that vhci doesn't support SG and sends the request to stub driver by splitting the SG list into multiple URBs, stub driver allocates a buffer for each URB with kmalloc() as it did before this patch. * Test environment Test uses two difference machines and two different kernel version to make mismatch situation between the client and the server where vhci supports SG, but stub driver does not, or vice versa. All tests are conducted in both full SG support that both vhci and stub support SG and half SG support that is the mismatch situation. Test kernel version is 5.3-rc6 with commit "usb: add a HCD_DMA flag instead of guestimating DMA capabilities" to avoid unnecessary DMA mapping and unmapping. - Test kernel version - 5.3-rc6 with SG support - 5.1.20-200.fc29.x86_64 without SG support * SG support test - Test devices - Super-speed storage device - SanDisk Ultra USB 3.0 - High-speed storage device - SMI corporation USB 2.0 flash drive - Test description Test read and write operation of mass storage device that uses the BULK transfer. In test, the client reads and writes files whose size is over 1G and it works normally. * Regression test - Test devices - Super-speed device - Logitech Brio webcam - High-speed device - Logitech C920 HD Pro webcam - Full-speed device - Logitech bluetooth mouse - Britz BR-Orion speaker - Low-speed device - Logitech wired mouse - Test description Moving and click test for mouse. To test the webcam, use gnome-cheese. To test the speaker, play music and video on the client. All works normally. * VUDC compatibility test VUDC also works well with this patch. Tests are done with two USB gadget created by CONFIGFS USB gadget. Both use the BULK pipe. 1. Serial gadget 2. Mass storage gadget - Serial gadget test Serial gadget on the host sends and receives data using cat command on the /dev/ttyGS<N>. The client uses minicom to communicate with the serial gadget. - Mass storage gadget test After connecting the gadget with vhci, use "dd" to test read and write operation on the client side. Read - dd if=/dev/sd<N> iflag=direct of=/dev/null bs=1G count=1 Write - dd if=<my file path> iflag=direct of=/dev/sd<N> bs=1G count=1 Signed-off-by: Suwan Kim <suwan.kim027@gmail.com> Acked-by: Shuah khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20190828032741.12234-1-suwan.kim027@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-08-28 11:27:41 +08:00
if (urb->num_sgs) {
unsigned int copy = urb->actual_length;
int size;
for_each_sg(urb->sg, sg, urb->num_sgs, i) {
if (copy == 0)
break;
if (copy < sg->length)
size = copy;
else
size = sg->length;
iov[iovnum].iov_base = sg_virt(sg);
iov[iovnum].iov_len = size;
iovnum++;
copy -= size;
}
} else {
iov[iovnum].iov_base = urb->transfer_buffer;
iov[iovnum].iov_len = urb->actual_length;
iovnum++;
}
txsize += urb->actual_length;
} else if (usb_pipein(urb->pipe) &&
usb_pipetype(urb->pipe) == PIPE_ISOCHRONOUS) {
/*
* For isochronous packets: actual length is the sum of
* the actual length of the individual, packets, but as
* the packet offsets are not changed there will be
* padding between the packets. To optimally use the
* bandwidth the padding is not transmitted.
*/
int i;
for (i = 0; i < urb->number_of_packets; i++) {
iov[iovnum].iov_base = urb->transfer_buffer +
urb->iso_frame_desc[i].offset;
iov[iovnum].iov_len =
urb->iso_frame_desc[i].actual_length;
iovnum++;
txsize += urb->iso_frame_desc[i].actual_length;
}
if (txsize != sizeof(pdu_header) + urb->actual_length) {
dev_err(&sdev->udev->dev,
"actual length of urb %d does not match iso packet sizes %zu\n",
urb->actual_length,
txsize-sizeof(pdu_header));
kfree(iov);
usbip_event_add(&sdev->ud,
SDEV_EVENT_ERROR_TCP);
return -1;
}
}
/* 3. setup iso_packet_descriptor */
if (usb_pipetype(urb->pipe) == PIPE_ISOCHRONOUS) {
ssize_t len = 0;
iso_buffer = usbip_alloc_iso_desc_pdu(urb, &len);
if (!iso_buffer) {
usbip_event_add(&sdev->ud,
SDEV_EVENT_ERROR_MALLOC);
kfree(iov);
return -1;
}
iov[iovnum].iov_base = iso_buffer;
iov[iovnum].iov_len = len;
txsize += len;
iovnum++;
}
ret = kernel_sendmsg(sdev->ud.tcp_socket, &msg,
iov, iovnum, txsize);
if (ret != txsize) {
dev_err(&sdev->udev->dev,
"sendmsg failed!, retval %d for %zd\n",
ret, txsize);
kfree(iov);
kfree(iso_buffer);
usbip_event_add(&sdev->ud, SDEV_EVENT_ERROR_TCP);
return -1;
}
kfree(iov);
kfree(iso_buffer);
total_size += txsize;
}
spin_lock_irqsave(&sdev->priv_lock, flags);
list_for_each_entry_safe(priv, tmp, &sdev->priv_free, list) {
stub_free_priv_and_urb(priv);
}
spin_unlock_irqrestore(&sdev->priv_lock, flags);
return total_size;
}
static struct stub_unlink *dequeue_from_unlink_tx(struct stub_device *sdev)
{
unsigned long flags;
struct stub_unlink *unlink, *tmp;
spin_lock_irqsave(&sdev->priv_lock, flags);
list_for_each_entry_safe(unlink, tmp, &sdev->unlink_tx, list) {
list_move_tail(&unlink->list, &sdev->unlink_free);
spin_unlock_irqrestore(&sdev->priv_lock, flags);
return unlink;
}
spin_unlock_irqrestore(&sdev->priv_lock, flags);
return NULL;
}
static int stub_send_ret_unlink(struct stub_device *sdev)
{
unsigned long flags;
struct stub_unlink *unlink, *tmp;
struct msghdr msg;
struct kvec iov[1];
size_t txsize;
size_t total_size = 0;
while ((unlink = dequeue_from_unlink_tx(sdev)) != NULL) {
int ret;
struct usbip_header pdu_header;
txsize = 0;
memset(&pdu_header, 0, sizeof(pdu_header));
memset(&msg, 0, sizeof(msg));
memset(&iov, 0, sizeof(iov));
usbip_dbg_stub_tx("setup ret unlink %lu\n", unlink->seqnum);
/* 1. setup usbip_header */
setup_ret_unlink_pdu(&pdu_header, unlink);
usbip_header_correct_endian(&pdu_header, 1);
iov[0].iov_base = &pdu_header;
iov[0].iov_len = sizeof(pdu_header);
txsize += sizeof(pdu_header);
ret = kernel_sendmsg(sdev->ud.tcp_socket, &msg, iov,
1, txsize);
if (ret != txsize) {
dev_err(&sdev->udev->dev,
"sendmsg failed!, retval %d for %zd\n",
ret, txsize);
usbip_event_add(&sdev->ud, SDEV_EVENT_ERROR_TCP);
return -1;
}
usbip_dbg_stub_tx("send txdata\n");
total_size += txsize;
}
spin_lock_irqsave(&sdev->priv_lock, flags);
list_for_each_entry_safe(unlink, tmp, &sdev->unlink_free, list) {
list_del(&unlink->list);
kfree(unlink);
}
spin_unlock_irqrestore(&sdev->priv_lock, flags);
return total_size;
}
int stub_tx_loop(void *data)
{
struct usbip_device *ud = data;
struct stub_device *sdev = container_of(ud, struct stub_device, ud);
while (!kthread_should_stop()) {
if (usbip_event_happened(ud))
break;
/*
* send_ret_submit comes earlier than send_ret_unlink. stub_rx
* looks at only priv_init queue. If the completion of a URB is
* earlier than the receive of CMD_UNLINK, priv is moved to
* priv_tx queue and stub_rx does not find the target priv. In
* this case, vhci_rx receives the result of the submit request
* and then receives the result of the unlink request. The
* result of the submit is given back to the usbcore as the
* completion of the unlink request. The request of the
* unlink is ignored. This is ok because a driver who calls
* usb_unlink_urb() understands the unlink was too late by
* getting the status of the given-backed URB which has the
* status of usb_submit_urb().
*/
if (stub_send_ret_submit(sdev) < 0)
break;
if (stub_send_ret_unlink(sdev) < 0)
break;
wait_event_interruptible(sdev->tx_waitq,
(!list_empty(&sdev->priv_tx) ||
!list_empty(&sdev->unlink_tx) ||
kthread_should_stop()));
}
return 0;
}