1=========== 2EHCI driver 3=========== 4 527-Dec-2002 6 7The EHCI driver is used to talk to high speed USB 2.0 devices using 8USB 2.0-capable host controller hardware. The USB 2.0 standard is 9compatible with the USB 1.1 standard. It defines three transfer speeds: 10 11 - "High Speed" 480 Mbit/sec (60 MByte/sec) 12 - "Full Speed" 12 Mbit/sec (1.5 MByte/sec) 13 - "Low Speed" 1.5 Mbit/sec 14 15USB 1.1 only addressed full speed and low speed. High speed devices 16can be used on USB 1.1 systems, but they slow down to USB 1.1 speeds. 17 18USB 1.1 devices may also be used on USB 2.0 systems. When plugged 19into an EHCI controller, they are given to a USB 1.1 "companion" 20controller, which is a OHCI or UHCI controller as normally used with 21such devices. When USB 1.1 devices plug into USB 2.0 hubs, they 22interact with the EHCI controller through a "Transaction Translator" 23(TT) in the hub, which turns low or full speed transactions into 24high speed "split transactions" that don't waste transfer bandwidth. 25 26At this writing, this driver has been seen to work with implementations 27of EHCI from (in alphabetical order): Intel, NEC, Philips, and VIA. 28Other EHCI implementations are becoming available from other vendors; 29you should expect this driver to work with them too. 30 31While usb-storage devices have been available since mid-2001 (working 32quite speedily on the 2.4 version of this driver), hubs have only 33been available since late 2001, and other kinds of high speed devices 34appear to be on hold until more systems come with USB 2.0 built-in. 35Such new systems have been available since early 2002, and became much 36more typical in the second half of 2002. 37 38Note that USB 2.0 support involves more than just EHCI. It requires 39other changes to the Linux-USB core APIs, including the hub driver, 40but those changes haven't needed to really change the basic "usbcore" 41APIs exposed to USB device drivers. 42 43- David Brownell 44 <dbrownell@users.sourceforge.net> 45 46 47Functionality 48============= 49 50This driver is regularly tested on x86 hardware, and has also been 51used on PPC hardware so big/little endianness issues should be gone. 52It's believed to do all the right PCI magic so that I/O works even on 53systems with interesting DMA mapping issues. 54 55Transfer Types 56-------------- 57 58At this writing the driver should comfortably handle all control, bulk, 59and interrupt transfers, including requests to USB 1.1 devices through 60transaction translators (TTs) in USB 2.0 hubs. But you may find bugs. 61 62High Speed Isochronous (ISO) transfer support is also functional, but 63at this writing no Linux drivers have been using that support. 64 65Full Speed Isochronous transfer support, through transaction translators, 66is not yet available. Note that split transaction support for ISO 67transfers can't share much code with the code for high speed ISO transfers, 68since EHCI represents these with a different data structure. So for now, 69most USB audio and video devices can't be connected to high speed buses. 70 71Driver Behavior 72--------------- 73 74Transfers of all types can be queued. This means that control transfers 75from a driver on one interface (or through usbfs) won't interfere with 76ones from another driver, and that interrupt transfers can use periods 77of one frame without risking data loss due to interrupt processing costs. 78 79The EHCI root hub code hands off USB 1.1 devices to its companion 80controller. This driver doesn't need to know anything about those 81drivers; a OHCI or UHCI driver that works already doesn't need to change 82just because the EHCI driver is also present. 83 84There are some issues with power management; suspend/resume doesn't 85behave quite right at the moment. 86 87Also, some shortcuts have been taken with the scheduling periodic 88transactions (interrupt and isochronous transfers). These place some 89limits on the number of periodic transactions that can be scheduled, 90and prevent use of polling intervals of less than one frame. 91 92 93Use by 94====== 95 96Assuming you have an EHCI controller (on a PCI card or motherboard) 97and have compiled this driver as a module, load this like:: 98 99 # modprobe ehci-hcd 100 101and remove it by:: 102 103 # rmmod ehci-hcd 104 105You should also have a driver for a "companion controller", such as 106"ohci-hcd" or "uhci-hcd". In case of any trouble with the EHCI driver, 107remove its module and then the driver for that companion controller will 108take over (at lower speed) all the devices that were previously handled 109by the EHCI driver. 110 111Module parameters (pass to "modprobe") include: 112 113 log2_irq_thresh (default 0): 114 Log2 of default interrupt delay, in microframes. The default 115 value is 0, indicating 1 microframe (125 usec). Maximum value 116 is 6, indicating 2^6 = 64 microframes. This controls how often 117 the EHCI controller can issue interrupts. 118 119If you're using this driver on a 2.5 kernel, and you've enabled USB 120debugging support, you'll see three files in the "sysfs" directory for 121any EHCI controller: 122 123 "async" 124 dumps the asynchronous schedule, used for control 125 and bulk transfers. Shows each active qh and the qtds 126 pending, usually one qtd per urb. (Look at it with 127 usb-storage doing disk I/O; watch the request queues!) 128 "periodic" 129 dumps the periodic schedule, used for interrupt 130 and isochronous transfers. Doesn't show qtds. 131 "registers" 132 show controller register state, and 133 134The contents of those files can help identify driver problems. 135 136 137Device drivers shouldn't care whether they're running over EHCI or not, 138but they may want to check for "usb_device->speed == USB_SPEED_HIGH". 139High speed devices can do things that full speed (or low speed) ones 140can't, such as "high bandwidth" periodic (interrupt or ISO) transfers. 141Also, some values in device descriptors (such as polling intervals for 142periodic transfers) use different encodings when operating at high speed. 143 144However, do make a point of testing device drivers through USB 2.0 hubs. 145Those hubs report some failures, such as disconnections, differently when 146transaction translators are in use; some drivers have been seen to behave 147badly when they see different faults than OHCI or UHCI report. 148 149 150Performance 151=========== 152 153USB 2.0 throughput is gated by two main factors: how fast the host 154controller can process requests, and how fast devices can respond to 155them. The 480 Mbit/sec "raw transfer rate" is obeyed by all devices, 156but aggregate throughput is also affected by issues like delays between 157individual high speed packets, driver intelligence, and of course the 158overall system load. Latency is also a performance concern. 159 160Bulk transfers are most often used where throughput is an issue. It's 161good to keep in mind that bulk transfers are always in 512 byte packets, 162and at most 13 of those fit into one USB 2.0 microframe. Eight USB 2.0 163microframes fit in a USB 1.1 frame; a microframe is 1 msec/8 = 125 usec. 164 165So more than 50 MByte/sec is available for bulk transfers, when both 166hardware and device driver software allow it. Periodic transfer modes 167(isochronous and interrupt) allow the larger packet sizes which let you 168approach the quoted 480 MBit/sec transfer rate. 169 170Hardware Performance 171-------------------- 172 173At this writing, individual USB 2.0 devices tend to max out at around 17420 MByte/sec transfer rates. This is of course subject to change; 175and some devices now go faster, while others go slower. 176 177The first NEC implementation of EHCI seems to have a hardware bottleneck 178at around 28 MByte/sec aggregate transfer rate. While this is clearly 179enough for a single device at 20 MByte/sec, putting three such devices 180onto one bus does not get you 60 MByte/sec. The issue appears to be 181that the controller hardware won't do concurrent USB and PCI access, 182so that it's only trying six (or maybe seven) USB transactions each 183microframe rather than thirteen. (Seems like a reasonable trade off 184for a product that beat all the others to market by over a year!) 185 186It's expected that newer implementations will better this, throwing 187more silicon real estate at the problem so that new motherboard chip 188sets will get closer to that 60 MByte/sec target. That includes an 189updated implementation from NEC, as well as other vendors' silicon. 190 191There's a minimum latency of one microframe (125 usec) for the host 192to receive interrupts from the EHCI controller indicating completion 193of requests. That latency is tunable; there's a module option. By 194default ehci-hcd driver uses the minimum latency, which means that if 195you issue a control or bulk request you can often expect to learn that 196it completed in less than 250 usec (depending on transfer size). 197 198Software Performance 199-------------------- 200 201To get even 20 MByte/sec transfer rates, Linux-USB device drivers will 202need to keep the EHCI queue full. That means issuing large requests, 203or using bulk queuing if a series of small requests needs to be issued. 204When drivers don't do that, their performance results will show it. 205 206In typical situations, a usb_bulk_msg() loop writing out 4 KB chunks is 207going to waste more than half the USB 2.0 bandwidth. Delays between the 208I/O completion and the driver issuing the next request will take longer 209than the I/O. If that same loop used 16 KB chunks, it'd be better; a 210sequence of 128 KB chunks would waste a lot less. 211 212But rather than depending on such large I/O buffers to make synchronous 213I/O be efficient, it's better to just queue up several (bulk) requests 214to the HC, and wait for them all to complete (or be canceled on error). 215Such URB queuing should work with all the USB 1.1 HC drivers too. 216 217In the Linux 2.5 kernels, new usb_sg_*() api calls have been defined; they 218queue all the buffers from a scatterlist. They also use scatterlist DMA 219mapping (which might apply an IOMMU) and IRQ reduction, all of which will 220help make high speed transfers run as fast as they can. 221 222 223TBD: 224 Interrupt and ISO transfer performance issues. Those periodic 225 transfers are fully scheduled, so the main issue is likely to be how 226 to trigger "high bandwidth" modes. 227 228TBD: 229 More than standard 80% periodic bandwidth allocation is possible 230 through sysfs uframe_periodic_max parameter. Describe that. 231