RE: [PATCH v3 00/36] thunderbolt: Software connection manager improvements

[Mario.Limonciello]

> -----Original Message-----
> From: Mika Westerberg <mika.westerberg@xxxxxxxxxxxxxxx>
> Sent: Thursday, March 28, 2019 7:36 AM
> To: linux-kernel@xxxxxxxxxxxxxxx
> Cc: Michael Jamet; Yehezkel Bernat; Andreas Noever; Lukas Wunner; David S .
> Miller; Andy Shevchenko; Christian Kellner; Limonciello, Mario; Mika Westerberg;
> netdev@xxxxxxxxxxxxxxx
> Subject: [PATCH v3 00/36] thunderbolt: Software connection manager
> improvements
> 
> 
> [EXTERNAL EMAIL]
> 
> Hi,
> 
> This is third iteration of the patch series intending to bring same kind of
> functionality for older Apple systems than we have in PCs. Software
> connection manager is used on Apple hardware with Light Ridge, Cactus Ridge
> or Falcon Ridge controllers to create PCIe tunnels when a Thunderbolt
> device is connected. Currently only one PCIe tunnel is supported. On newer
> Alpine Ridge based Apple systems the driver starts the firmware which then
> takes care creating tunnels.
> 
> This series improves the software connection manager so that it will
> support:
> 
>   - Full PCIe daisy chains (up to 6 devices)
>   - Display Port tunneling
>   - P2P networking
> 
> We also add support for Titan Ridge based Apple systems where we can use
> the same flows than with Alpine Ridge to start the firmware.

It seems to me that there would be an expectation that PC system firmware and TBT controller
firmware is configured to behave like Apple systems to use this SW connection manager
instead of the ICM in AR/TR FW.

Is there an intent to eventually offer a way to "side-step" the TBT ICM and try to use this instead
without firmware support?

> 
> This applies on top of thunderbolt.git/next.
> 
> Christian, Mario do you see any issues with patch [05/36] regarding bolt
> and fwupd? The kernel is supposed to restart the syscall automatically so
> userspace should not be affected but wanted to check with you.

I don't see a problem for fwupd in this area.

> 
> Previous version of the patch series can be viewed here:
> 
>   v2: https://lkml.org/lkml/2019/2/6/347
>   v1: https://lkml.org/lkml/2019/1/29/924
> 
> Making v3 took longer than I anticipated mostly due to some issues I run
> during testing the new changes. There are quite many changes so I dropped
> the reviewed-by tags I got for v2. Below is the list of major changes from
> the previous version:
> 
>   * Always set port->remote even in case of dual link connection.
> 
>   * Leave (DP, PCIe) tunnels up when the driver is unloaded. When loaded
>     back, it discovers the existing tunnels and updated data structures
>     accordingly. I noticed that the code in v2 did not support cases
>     properly when you unplug something before the driver gets loaded back.
>     This version tears down partial paths during discovery.
> 
>   * Do not automatically create PCIe tunnels. Instead we implement "user"
>     security level in the software connection manager as well taking
>     advantage of the existing sysfs interfaces. This allows user to disable
>     PCIe tunneling completely or implement different white listing
>     policies. Major distros include bolt system daemon that takes care of
>     this.

This is a bit unfortunate.  Is this because of IOMMU limitations in working
with devices down the chain?

> 
>   * When testing on two-port Falcon Ridge based system I realized
>     that we always just pick the first available PCIe downstream adapter
>     regardless of which Thunderbolt port you plug the device which is not
>     consistent. To solve this we add mapping between host PCIe downstream
>     adapter and the Thunderbolt port for Cactus Ridge and Falcon Ridge
>     based systems.
> 
>   * Take domain lock in switch sysfs callbacks. This is needed because the
>     software connection manager needs to walk over the topology during
>     tunnel creation so switch_lock is not enough anymore.
> 
> Changes from v1:
> 
>   * Added ACK from David
> 
>   * Add constant (TMU_ACCESS_EN) for BIT(20) when TMU access is enabled. We
>     keep it in cap.c close to the LR/ER workaround. Also we enable/disable
>     only during capability walk. If it turns we need to have it enabled
>     elsewhere we can move it to switch.c and enable just once during
>     switch enumeration.
> 
>   * Use 0 to mean no cap_adap instead of negative value. This follows
>     cap_phy.
> 
>   * Use correct PCI IDs (_BRIDGE) in the last patch where we start firmware
>     on Titan Ridge. It wrongly used NHI PCI IDs in v1.
> 
> Mika Westerberg (36):
>   net: thunderbolt: Unregister ThunderboltIP protocol handler when suspending
>   thunderbolt: Remove unused work field in struct tb_switch
>   thunderbolt: Drop duplicated get_switch_by_route()
>   thunderbolt: Block reads and writes if switch is unplugged
>   thunderbolt: Take domain lock in switch sysfs attribute callbacks
>   thunderbolt: Do not allocate switch if depth is greater than 6
>   thunderbolt: Enable TMU access when accessing port space on legacy devices
>   thunderbolt: Add dummy read after port capability list walk on Light Ridge
>   thunderbolt: Move LC specific functionality into a separate file
>   thunderbolt: Configure lanes when switch is initialized
>   thunderbolt: Set sleep bit when suspending switch
>   thunderbolt: Properly disable path
>   thunderbolt: Cache adapter specific capability offset into struct port
>   thunderbolt: Rename tunnel_pci to tunnel
>   thunderbolt: Generalize tunnel creation functionality
>   thunderbolt: Add functions for allocating and releasing HopIDs
>   thunderbolt: Assign remote for both ports in case of dual link
>   thunderbolt: Add helper function to iterate from one port to another
>   thunderbolt: Extend tunnel creation to more than 2 adjacent switches
>   thunderbolt: Deactivate all paths before restarting them
>   thunderbolt: Discover preboot PCIe paths the boot firmware established
>   thunderbolt: Add support for full PCIe daisy chains
>   thunderbolt: Scan only valid NULL adapter ports in hotplug
>   thunderbolt: Generalize port finding routines to support all port types
>   thunderbolt: Rework NFC credits handling
>   thunderbolt: Add support for Display Port tunnels
>   thunderbolt: Do not tear down tunnels when driver is unloaded
>   thunderbolt: Run tb_xdp_handle_request() in system workqueue
>   thunderbolt: Add XDomain UUID exchange support
>   thunderbolt: Add support for DMA tunnels
>   thunderbolt: Make tb_switch_alloc() return ERR_PTR()
>   thunderbolt: Add support for XDomain connections
>   thunderbolt: Make __TB_[SW|PORT]_PRINT take const parameters
>   thunderbolt: Make rest of the logging to happen at debug level
>   thunderbolt: Reword output of tb_dump_hop()
>   thunderbolt: Start firmware on Titan Ridge Apple systems
> 
>  drivers/net/thunderbolt.c        |   3 +
>  drivers/thunderbolt/Makefile     |   4 +-
>  drivers/thunderbolt/cap.c        |  85 +++-
>  drivers/thunderbolt/ctl.c        |   2 +-
>  drivers/thunderbolt/icm.c        |  60 ++-
>  drivers/thunderbolt/lc.c         | 179 ++++++++
>  drivers/thunderbolt/nhi.c        |   3 +-
>  drivers/thunderbolt/path.c       | 421 ++++++++++++++++---
>  drivers/thunderbolt/switch.c     | 551 +++++++++++++++++++-----
>  drivers/thunderbolt/tb.c         | 608 ++++++++++++++++++++-------
>  drivers/thunderbolt/tb.h         | 173 +++++++-
>  drivers/thunderbolt/tb_msgs.h    |  11 +
>  drivers/thunderbolt/tb_regs.h    |  50 ++-
>  drivers/thunderbolt/tunnel.c     | 691 +++++++++++++++++++++++++++++++
>  drivers/thunderbolt/tunnel.h     |  78 ++++
>  drivers/thunderbolt/tunnel_pci.c | 226 ----------
>  drivers/thunderbolt/tunnel_pci.h |  31 --
>  drivers/thunderbolt/xdomain.c    | 147 ++++++-
>  include/linux/thunderbolt.h      |   8 +
>  19 files changed, 2680 insertions(+), 651 deletions(-)
>  create mode 100644 drivers/thunderbolt/lc.c
>  create mode 100644 drivers/thunderbolt/tunnel.c
>  create mode 100644 drivers/thunderbolt/tunnel.h
>  delete mode 100644 drivers/thunderbolt/tunnel_pci.c
>  delete mode 100644 drivers/thunderbolt/tunnel_pci.h
> 
> --
> 2.20.1