281a193b53
configure sa(4) to request no I/O splitting by default. For tape devices, the user needs to be able to clearly understand what blocksize is actually being used when writing to a tape device. The previous behavior of physio(9) was that it would split up any I/O that was too large for the device, or too large to fit into MAXPHYS. This means that if, for instance, the user wrote a 1MB block to a tape device, and MAXPHYS was 128KB, the 1MB write would be split into 8 128K chunks. This would be done without informing the user. This has suboptimal effects, especially when trying to communicate status to the user. In the event of an error writing to a tape (e.g. physical end of tape) in the middle of a 1MB block that has been split into 8 pieces, the user could have the first two 128K pieces written successfully, the third returned with an error, and the last 5 returned with 0 bytes written. If the user is using a standard write(2) system call, all he will see is the ENOSPC error. He won't have a clue how much actually got written. (With a writev(2) system call, he should be able to determine how much got written in addition to the error.) The solution is to prevent physio(9) from splitting the I/O. The new cdev flag, SI_NOSPLIT, tells physio that the driver does not want I/O to be split beforehand. Although the sa(4) driver now enables SI_NOSPLIT by default, that can be disabled by two loader tunables for now. It will not be configurable starting in FreeBSD 11.0. kern.cam.sa.allow_io_split allows the user to configure I/O splitting for all sa(4) driver instances. kern.cam.sa.%d.allow_io_split allows the user to configure I/O splitting for a specific sa(4) instance. There are also now three sa(4) driver sysctl variables that let the users see some sa(4) driver values. kern.cam.sa.%d.allow_io_split shows whether I/O splitting is turned on. kern.cam.sa.%d.maxio shows the maximum I/O size allowed by kernel configuration parameters (e.g. MAXPHYS, DFLTPHYS) and the capabilities of the controller. kern.cam.sa.%d.cpi_maxio shows the maximum I/O size supported by the controller. Note that a better long term solution would be to implement support for chaining buffers, so that that MAXPHYS is no longer a limiting factor for I/O size to tape and disk devices. At that point, the controller and the tape drive would become the limiting factors. sys/conf.h: Add a new cdev flag, SI_NOSPLIT, that allows a driver to tell physio not to split up I/O. sys/param.h: Bump __FreeBSD_version to 1000049 for the addition of the SI_NOSPLIT cdev flag. kern_physio.c: If the SI_NOSPLIT flag is set on the cdev, return any I/O that is larger than si_iosize_max or MAXPHYS, has more than one segment, or would have to be split because of misalignment with EFBIG. (File too large). In the event of an error, print a console message to give the user a clue about what happened. scsi_sa.c: Set the SI_NOSPLIT cdev flag on the devices created for the sa(4) driver by default. Add tunables to control whether we allow I/O splitting in physio(9). Explain in the comments that allowing I/O splitting will be deprecated for the sa(4) driver in FreeBSD 11.0. Add sysctl variables to display the maximum I/O size we can do (which could be further limited by read block limits) and the maximum I/O size that the controller can do. Limit our maximum I/O size (recorded in the cdev's si_iosize_max) by MAXPHYS. This isn't strictly necessary, because physio(9) will limit it to MAXPHYS, but it will provide some clarity for the application. Record the controller's maximum I/O size reported in the Path Inquiry CCB. sa.4: Document the block size behavior, and explain that the option of allowing physio(9) to split the I/O will disappear in FreeBSD 11.0. Sponsored by: Spectra Logic |
||
---|---|---|
.. | ||
ata | ||
ctl | ||
scsi | ||
cam_ccb.h | ||
cam_compat.c | ||
cam_compat.h | ||
cam_debug.h | ||
cam_periph.c | ||
cam_periph.h | ||
cam_queue.c | ||
cam_queue.h | ||
cam_sim.c | ||
cam_sim.h | ||
cam_xpt_internal.h | ||
cam_xpt_periph.h | ||
cam_xpt_sim.h | ||
cam_xpt.c | ||
cam_xpt.h | ||
cam.c | ||
cam.h | ||
README.quirks |
/* $FreeBSD$ */ FreeBSD Quirk Guidelines Nate Lawson - njl at freebsd org 0. Introduction FreeBSD drivers make every attempt possible to support the standards behind hardware. Where possible and not in conflict with the standard, they also attempt to work around hardware which doesn't strictly conform. However, some devices have flaws which can't be worked around while keeping the driver compatible with the standard. For these devices, we have created a quirks mechanism to indicate to the driver that it must avoid certain commands or use them differently with a specific model and/or version of hardware. This document focuses on identifying and committing quirks for storage hardware involving CAM and UMASS but is applicable to other areas. CAM provides a generic transport for SCSI-like devices. Many different transports use SCSI command sets including parallel SCSI, firewire (1394), USB UMASS, fibre channel, and ATAPI. For block devices (i.e. hard drives, flash adapters, cameras) there are two standards, SBC and RBC. SCSI hard drives are usually SBC-compliant and smaller devices like flash drives are usually RBC-compliant. Multimedia devices including CDROMs and DVD-RW are usually MMC-compliant. Please follow these guidelines to get your device working as soon as possible. If you are a committer, please do NOT commit quirks directly but follow this process also. 1. Determing the problem The first step is to determine what's wrong. If the device should be supported but hangs while attaching, it's possible a quirk can help. The types of things a quirk can fix are: ` * cam/cam_xpt.c quirks o CAM_QUIRK_NOLUNS - do not probe luns other than 0 since device responds to all inquiries with "lun present". o CAM_QUIRK_NOSERIAL - do not send an inquiry for serial number. o CAM_QUIRK_HILUNS - probe all luns even if some respond "not present" since device has a sparse lun space. * cam/scsi/scsi_da.c quirks o DA_Q_NO_SYNC_CACHE - The sync cache command is used to force a drive to write out all changes to disk before shutting down. Some drives hang when receiving this command even though it is required by all SBC and RBC standards. Note that a warning message on console is NOT sufficient to add this quirk. The warning messages are harmless and only a device or system hang is cause for adding this quirk. o DA_Q_NO_6_BYTE - The RBC spec (see Links below) does not allow for 6-byte READ/WRITE commands. Some manufacturers took that too literally and crash when receiving 6-byte commands. This quirk causes FreeBSD to only send 10-byte commands. Since the CAM subsystem has been modified to not send 6-byte commands to USB, 1394, and other transports that don't support SBC, this quirk should be very rare. o DA_Q_NO_PREVENT - Don't use the prevent/allow commands to keep a removable medium from being ejected. Some systems can't handle these commands (rare). * cam/scsi/scsi_cd.c quirks o CD_Q_NO_TOUCH - not implemented o CD_Q_BCD_TRACKS - convert start/end track to BCD o CD_Q_NO_CHANGER - never treat as a changer o CD_Q_CHANGER - always treat as a changer * cam/scsi/scsi_ch.c quirks o CH_Q_NO_DBD - disable block descriptors in mode sense * cam/scsi/scsi_sa.c quirks o SA_QUIRK_NOCOMP - Can't deal with compression at all o SA_QUIRK_FIXED - Force fixed mode o SA_QUIRK_VARIABLE - Force variable mode o SA_QUIRK_2FM - Needs Two File Marks at EOD o SA_QUIRK_1FM - No more than 1 File Mark at EOD o SA_QUIRK_NODREAD - Don't try and dummy read density o SA_QUIRK_NO_MODESEL - Don't do mode select at all o SA_QUIRK_NO_CPAGE - Don't use DEVICE COMPRESSION page * dev/usb/umass.c quirks o NO_TEST_UNIT_READY - The drive does not support Test Unit Ready. Convert to Start Unit. This command is a simple no-op for most firmware but some of them hang when this command is sent. o RS_NO_CLEAR_UA - The drive does not reset the Unit Attention state after REQUEST SENSE has been sent. The INQUIRY command does not reset the UA either, and so CAM runs in circles trying to retrieve the initial INQUIRY data. This quirk signifies that after a unit attention condition, don't try to clear the condition with a request sense command. o NO_START_STOP - Like test unit ready, don't send this command if it hangs the device. o FORCE_SHORT_INQUIRY - Don't ask for full inquiry data (256 bytes). Some drives can only handle the shorter inquiry length (36 bytes). o SHUTTLE_INIT - Needs to be initialised the Shuttle way. Haven't looked into what this does but apparently it's mostly Shuttle devices. o ALT_IFACE_1 - Drive needs to be switched to alternate interface 1. Rare. o FLOPPY_SPEED - Drive does not do 1Mb/s, but just floppy speeds (20kb/s). o IGNORE_RESIDUE - The device can't count and gets the residue of transfers wrong. This is sometimes needed for devices where large transfers cause stalls. o NO_GETMAXLUN - Get maximum LUN is a command to identify multiple devices sharing the same ID. For instance, a multislot compact flash reader might be on two LUNS. Some non-standard devices hang when receiving this command so this quirk disables it. o WRONG_CSWSIG - The device uses a weird CSWSIGNATURE. Rare. o NO_INQUIRY - Device cannot handle INQUIRY so fake a generic response. INQUIRY is one of the most basic commands but some drives can't even handle it. (No idea how such devices even work at all on other OS's.) This quirk fakes up a valid but generic response for devices that can't handle INQUIRY. o NO_INQUIRY_EVPD - Device cannot handle an extended INQUIRY asking for vital product data (EVPD) so just return a "no data" response (check condition) without sending the command to the device. 2. Testing a Quirk After you have an idea what you want to try, edit the proper file above, using wildcarding to be sure your device is matched. Here is a list of the common things to try. Note that some devices require multiple quirks or quirks in different drivers. For example, some USB pen drives or flash readers require quirks in both da(4) and umass(4). * umass(4) device (sys/dev/usb/umass.c) -- this quirk matches an Asahi Optical device with any product ID or revision ID. * * { USB_VENDOR_ASAHIOPTICAL, PID_WILDCARD, RID_WILDCARD, * UMASS_PROTO_ATAPI | UMASS_PROTO_CBI_I, * RS_NO_CLEAR_UA * }, * da(4) device (sys/cam/scsi/scsi_da.c) -- this quirk matches a Creative device with a name of "NOMAD_MUVO" and any revision. * * { * /* * * Creative Nomad MUVO mp3 player (USB) * * PR: kern/53094 * */ * {T_DIRECT, SIP_MEDIA_REMOVABLE, "CREATIVE", "NOMAD_MUVO", "*"}, * /*quirks*/ DA_Q_NO_SYNC_CACHE|DA_Q_NO_PREVENT * }, 3. Filing a PR All quirk submissions MUST go through GNATS. For information on how to submit a PR, see this page. Please include the following in your PR: * Subject: QUIRK: FooCo USB DVD-RAM drive * Output of "camcontrol inquiry yourdevice" * Manufacturer name, model number, etc. * Transport type (FC, SCSI, USB, Firewire) * Output from dmesg for failed attach attempts * Output from dmesg for successful attach attempts (after quirk added) * Output of "usbdevs -v" with device attached * Valid email address Here are some examples of well-formed PRs: * kern/43580 * kern/49054 4. What happens next I will review your submission, respond with comments, and once the quirk is deemed necessary and ready for committing, I'll commit it, referencing the PR. (Again, all quirks must be submitted as PRs). Questions? Email njl AT freebsd.org. 5. Note to Committers Please insert quirks in the right section in scsi_da.c, sorted by PR number. Always include the name and PR number for scsi_da.c (see above for an example.) Please sort quirks alphabetically in umass.c. Follow the surrounding style in all drivers. Be sure to correspond with the submitter to be sure the quirk you are adding is the minimum necessary, not quirking other useful features and not overly broad (i.e., too many wildcards).