When chaining ISOC TRBs together, only the first ISOC TRB should be of
type ISOC_FIRST, all others should be of type ISOC. This patch fixes
that.
Fixes: c6267a51639b ("usb: dwc3: gadget: align transfers to wMaxPacketSize")
Cc: # v4.11+
Signed-off-by: Felipe Balbi
---
drivers/usb/dwc3/gadget.c |
Hi,
Thinh Nguyen writes:
> On 8/2/2018 12:46 AM, Felipe Balbi wrote:
>> Hi,
>>
>> Thinh Nguyen writes:
>>
>>
>>
These patches will not fix the issue.
>>> What do you think of this fix?
>>>
>>> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
>>> index f452ab708ffc..
Existing code is not applicable to big-endian machines
ctrlrequest fields received in USB endian - i.e. in little-endian
and should be converted to cpu endianness before usage.
Signed-off-by: Alexey Spirkov
---
drivers/usb/musb/musb_gadget_ep0.c | 29 ++---
1 file changed
On Fri, Aug 3, 2018 at 12:03 PM, Alexey Spirkov wrote:
> Existing code is not applicable to big-endian machines
> ctrlrequest fields received in USB endian - i.e. in little-endian
> and should be converted to cpu endianness before usage.
> - epnum = (u8) ctrlrequest->wIndex;
> +
Add parameter 'tag' to function usbtmc_ioctl_abort_bulk_in_tag()
for future versions.
Remove calculation of max_size (=wMaxPacketSize) and wrong
condition (actual == max_size) in while loop. An abort operation
should always flush the complete Bulk-IN until a short packet is
received.
Return error
The new ioctl USBTMC_IOCTL_WRITE sends a generic message to bulk OUT.
This ioctl is used for vendor specific or asynchronous I/O as well.
The message is split into chunks of 4k (page size).
Message size is aligned to 32 bit boundaries.
With flag USBTMC_FLAG_ASYNC the ioctl is non blocking.
With f
Add USBTMC_IOCTL_CTRL_REQUEST to send arbitrary requests on the
control pipe. Used by specific applications of IVI Foundation,
Inc. to implement VISA API functions: viUsbControlIn/Out.
The maximum length of control request is set to 4k.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
--
Submitted urbs are not allowed when system is suspended.
Thus the submitted urb waiting at interrupt pipe is killed
during suspend callback and submitted again when system resumes.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 16 +++-
1 fil
Add parameter 'tag' to function usbtmc_ioctl_abort_bulk_out_tag()
for future versions.
Use USBTMC_BUFSIZE (4k) instead of USBTMC_SIZE_IOBUFFER (2k).
Using USBTMC_SIZE_IOBUFFER is deprecated.
Insert a sleep of 50 ms between subsequent
CHECK_ABORT_BULK_OUT_STATUS control requests to avoid stressing
ioctl USBTMC_IOCTL_CANCEL_IO stops and kills all flying urbs of
last USBTMC_IOCTL_READ and USBTMC_IOCTL_WRITE function calls.
A subsequent call to USBTMC_IOCTL_READ or
USBTMC_IOCTL_WRITE_RESULT returns -ECANCELED with
information about current transferred data.
Signed-off-by: Guido Kiener
Reviewe
Add ioctl USBTMC_IOCTL_AUTO_ABORT to configure auto_abort for
each specific file handle.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 23 ---
include/uapi/linux/usb/tmc.h | 1 +
2 files changed, 17 insertions(+), 7 deletions(-)
add ioctl USBTMC_IOCTL_MSG_IN_ATTR that returns the specific
bmTransferAttributes field of the last DEV_DEP_MSG_IN Bulk-IN
header. This header is received by the read() function. The
meaning of the (u8) bitmap bmTransferAttributes is:
Bit 0 = EOM flag is set when the last transfer of a USBTMC
mess
Use new usbtmc_generic_write function to maximize bandwidth
during long data transfer.
The maximum output transfer size is limited to INT_MAX (=2GB).
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 176 +++--
1 file changed
ioctl USBTMC_IOCTL_WRITE_RESULT copies current out_transfer_size
to given __u32 pointer and returns current out_status of the last
(asnynchronous) USBTMC_IOCTL_WRITE call.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 25 +
inc
Remove calculation of max_size (=wMaxPacketSize) and wrong
condition (actual == max_size) in while loop. A device clear
should always flush the complete Bulk-IN FIFO.
Insert a sleep of 50 ms between subsequent CHECK_CLEAR_STATUS
control requests to avoid stressing the instrument with
repeated requ
Use new usbtmc_generic_read function to maximize bandwidth
during long data transfer. Also fix reading of zero length
packet (ZLP) or trailing short packet.
The maximum input transfer size is limited to INT_MAX (=2GB).
Also remove redundant return in send_request_dev_dep_msg_in().
Signed-off-by: G
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c
index 785eee3a6b89..6aef3447adcc 100644
--- a/drivers/usb/class/usbtmc.c
+++ b/dri
As all the properties of the usbtmc driver can now be
controlled on a per file descriptor basis by ioctl functions
the sysfs interface is of limited use.
We are not aware about applications that are using the sysfs
parameter TermChar, TermCharEnabled or auto_abort.
Signed-off-by: Guido Kiener
Rev
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 6 --
1 file changed, 6 deletions(-)
diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c
index 81144c0c4772..785eee3a6b89 100644
--- a/drivers/usb/class/usbtmc.c
+++ b/drivers/usb/class/
Remove redundant code and fix debug messages.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 25 -
1 file changed, 8 insertions(+), 17 deletions(-)
diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c
index fb49a7
Reserve a suitable range of ioctl numbers for USBTMC driver.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
Documentation/ioctl/ioctl-number.txt | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/Documentation/ioctl/ioctl-number.txt
b/Documentation/ioctl/ioctl-numb
The working group "VISA for Linux" of the IVI Foundation
www.ivifoundation.org specifies common rules, shared libraries and
drivers to implement the specification of "VPP-4.3: The VISA Library"
on Linux to be compatible with implementations on other operating systems.
The USBTMC protocol is part o
Wait until an SRQ (service request) is received on the interrupt pipe
or until the given period of time is expired. In contrast to the
poll() function this ioctl does not return when other (a)synchronous
I/O operations fail with EPOLLERR.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
--
Add ioctl USBTMC_IOCTL_API_VERSION to get current API version
of usbtmc driver.
This is to allow an instrument library to determine whether
the driver API is compatible with the implementation.
The API may change in future versions. Therefore the macro
USBTMC_API_VERSION should be incremented whe
Use common timeout macro USB_CTRL_GET_TIMEOUT (=5s) for all
usb_control_msg() function calls.
The macro USBTMC_TIMEOUT should only be used as default value for
Bulk IN/OUT transfers.
Signed-off-by: Guido Kiener
Reviewed-by: Steve Bayless
---
drivers/usb/class/usbtmc.c | 8
1 file chan
The USBTMC_IOCTL_READ call provides for generic synchronous and
asynchronous reads on bulk IN to implement vendor specific library
routines.
Depending on transfer_size the function submits one or more urbs (up
to 16) each with a size of up to 4kB.
The flag USBTMC_FLAG_IGNORE_TRAILER can be used w
The ioctl USBTMC_IOCTL_CLEANUP_IO kills all submitted urbs to OUT
and IN bulk, and clears all received data from IN bulk. Internal
transfer counters and error states are reset.
An application should use this ioctl after an asnychronous transfer
was canceled and/or error handling has finished.
Sig
Hi Andy,
Before my change:
CHECK drivers/usb/musb/musb_gadget_ep0.c
drivers/usb/musb/musb_gadget_ep0.c:85:26: warning: cast from restricted __le16
drivers/usb/musb/musb_gadget_ep0.c:220:58: warning: restricted __le16 degrades
to integer
drivers/usb/musb/musb_gadget_ep0.c:227:48: warning: rest
Hi,
Please don't top-posting.
On Fri, Aug 03, 2018 at 06:25:50AM +, Alexey Spirkov wrote:
> Hi, Bin,
>
> Thanks for note, but what do you think - is 'dma_mapping_error'
> checking still needed before call of unmap_dma_buffer? If yes than
> I'll check DMA usage in this check instead of is_buf
Hi,
I've run into a huge performance gap between running a g_mass_storage
gadget with the backing file pointing to a raw partition (/dev/mmcblk1)
vs. creating a file on an ext4 filesystem and exporting that.
With the partition I see write rates around 5MB/s, and with
/proc/sys/vm/block_dump
en
The drive (128GB USB 3.0)
https://www.sandisk.co.uk/home/usb-flash/ultra-fit-usb overheats to
the point where the metal part would burn someone if they held it for
more than a second.
The overheating only happens on this brand of USB drives, other USB
3.0 drives have been fine.
The overheating onl
On Fri, 3 Aug 2018, Martin Hicks wrote:
>
> Hi,
>
> I've run into a huge performance gap between running a g_mass_storage
> gadget with the backing file pointing to a raw partition (/dev/mmcblk1)
> vs. creating a file on an ext4 filesystem and exporting that.
>
> With the partition I see write
On Fri, Aug 03, 2018 at 04:15:22PM +0100, Mustafa A wrote:
> The drive (128GB USB 3.0)
> https://www.sandisk.co.uk/home/usb-flash/ultra-fit-usb overheats to
> the point where the metal part would burn someone if they held it for
> more than a second.
>
> The overheating only happens on this brand
33 matches
Mail list logo
