Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
切换分支/标签
  1. 23 5月, 2020 3 次提交
  3. 22 5月, 2020 17 次提交
  5. 15 5月, 2020 2 次提交
    • J
      xdp: For Intel AF_XDP drivers add XDP frame_sz · 2a637c5b
      由 Jesper Dangaard Brouer 提交于 
      Intel drivers implement native AF_XDP zerocopy in separate C-files,
that have its own invocation of bpf_prog_run_xdp(). The setup of
xdp_buff is also handled in separately from normal code path.

This patch update XDP frame_sz for AF_XDP zerocopy drivers i40e, ice
and ixgbe, as the code changes needed are very similar.  Introduce a
helper function xsk_umem_xdp_frame_sz() for calculating frame size.
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Acked-by: NBjörn Töpel <bjorn.topel@intel.com>
Cc: intel-wired-lan@lists.osuosl.org
Cc: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/158945347511.97035.8536753731329475655.stgit@firesoul
      2a637c5b
    • J
      ice: Add XDP frame size to driver · d4ecdbf7
      由 Jesper Dangaard Brouer 提交于 
      This driver uses different memory models depending on PAGE_SIZE at
compile time. For PAGE_SIZE 4K it uses page splitting, meaning for
normal MTU frame size is 2048 bytes (and headroom 192 bytes). For
larger MTUs the driver still use page splitting, by allocating
order-1 pages (8192 bytes) for RX frames. For PAGE_SIZE larger than
4K, driver instead advance its rx_buffer->page_offset with the frame
size "truesize".

For XDP frame size calculations, this mean that in PAGE_SIZE larger
than 4K mode the frame_sz change on a per packet basis. For the page
split 4K PAGE_SIZE mode, xdp.frame_sz is more constant and can be
updated once outside the main NAPI loop.

The default setting in the driver uses build_skb(), which provides
the necessary headroom and tailroom for XDP-redirect in RX-frame
(in both modes).

There is one complication, which is legacy-rx mode (configurable via
ethtool priv-flags). There are zero headroom in this mode, which is a
requirement for XDP-redirect to work. The conversion to xdp_frame
(convert_to_xdp_frame) will detect this insufficient space, and
xdp_do_redirect() call will fail. This is deemed acceptable, as it
allows other XDP actions to still work in legacy-mode. In
legacy-mode + larger PAGE_SIZE due to lacking tailroom, we also
accept that xdp_adjust_tail shrink doesn't work.
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Cc: intel-wired-lan@lists.osuosl.org
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: Alexander Duyck <alexander.duyck@gmail.com>
Link: https://lore.kernel.org/bpf/158945347002.97035.328088795813704587.stgit@firesoul
      d4ecdbf7
  7. 01 5月, 2020 1 次提交
  9. 29 3月, 2020 1 次提交
  11. 27 3月, 2020 1 次提交
  13. 21 3月, 2020 8 次提交
    • J
      ice: add board identifier info to devlink .info_get · e961b679
      由 Jacob Keller 提交于 
      Export a unique board identifier using "board.id" for devlink's
.info_get command.

Obtain this by reading the NVM for the PBA identification string.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Reviewed-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Acked-by: NJakub Kicinski <kuba@kernel.org>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      e961b679
    • J
      ice: add basic handler for devlink .info_get · ff2e5c70
      由 Jacob Keller 提交于 
      The devlink .info_get callback allows the driver to report detailed
version information. The following devlink versions are reported with
this initial implementation:

 "fw.mgmt" -> The version of the firmware that controls PHY, link, etc
 "fw.mgmt.api" -> API version of interface exposed over the AdminQ
 "fw.mgmt.build" -> Unique build id of the source for the management fw
 "fw.undi" -> Version of the Option ROM containing the UEFI driver
 "fw.psid.api" -> Version of the NVM image format.
 "fw.bundle_id" -> Unique identifier for the combined flash image.
 "fw.app.name" -> The name of the active DDP package.
 "fw.app" -> The version of the active DDP package.

With this, devlink dev info can report at least as much information as
is reported by ETHTOOL_GDRVINFO.

Compare the output from ethtool vs from devlink:

  $ ethtool -i ens785s0
  driver: ice
  version: 0.8.1-k
  firmware-version: 0.80 0x80002ec0 1.2581.0
  expansion-rom-version:
  bus-info: 0000:3b:00.0
  supports-statistics: yes
  supports-test: yes
  supports-eeprom-access: yes
  supports-register-dump: yes
  supports-priv-flags: yes

  $ devlink dev info pci/0000:3b:00.0
  pci/0000:3b:00.0:
  driver ice
  serial number 00-01-ab-ff-ff-ca-05-68
  versions:
      running:
        fw.mgmt 2.1.7
        fw.mgmt.api 1.5
        fw.mgmt.build 0x305d955f
        fw.undi 1.2581.0
        fw.psid.api 0.80
        fw.bundle_id 0x80002ec0
        fw.app.name ICE OS Default Package
        fw.app 1.3.1.0

More pieces of information can be displayed, each version is kept
separate instead of munged together, and each version has an identifier
which comes with associated documentation.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Acked-by: NJakub Kicinski <kuba@kernel.org>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      ff2e5c70
    • J
      ice: enable initial devlink support · 1adf7ead
      由 Jacob Keller 提交于 
      Begin implementing support for the devlink interface with the ice
driver.

The pf structure is currently memory managed through devres, via
a devm_alloc. To mimic this behavior, after allocating the devlink
pointer, use devm_add_action to add a teardown action for releasing the
devlink memory on exit.

The ice hardware is a multi-function PCIe device. Thus, each physical
function will get its own devlink instance. This means that each
function will be treated independently, with its own parameters and
configuration. This is done because the ice driver loads a separate
instance for each function.

Due to this, the implementation does not enable devlink to manage
device-wide resources or configuration, as each physical function will
be treated independently. This is done for simplicity, as managing
a devlink instance across multiple driver instances would significantly
increase the complexity for minimal gain.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Reviewed-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      1adf7ead
    • J
      ice: implement full NVM read from ETHTOOL_GEEPROM · 84a24798
      由 Jesse Brandeburg 提交于 
      The current implementation of .get_eeprom only enables reading from the
Shadow RAM portion of the NVM contents. Implement support for reading
the entire flash contents instead of only the initial portion contained
in the Shadow RAM.

A complete dump can take several seconds, but the ETHTOOL_GEEPROM ioctl
is capable of reading only a limited portion at a time by specifying the
offset and length to read.

In order to perform the reads directly, several functions are made non
static. Additionally, the unused ice_read_sr_buf_aq and ice_read_sr_buf
functions are removed.
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      84a24798
    • J
      ice: discover and store size of available flash · 81f07491
      由 Jacob Keller 提交于 
      When reading from the NVM using a flat address, it is useful to know the
upper bound on the size of the flash contents. This value is not stored
within the NVM.

We can determine the size by performing a bisection between upper and
lower bounds. It is known that the size cannot exceed 16 MB (offset of
0xFFFFFF).

Use a while loop to bisect the upper and lower bounds by reading one
byte at a time. On a failed read, lower the maximum bound. On
a successful read, increase the lower bound.

Save this as the flash_size in the ice_nvm_info structure that contains
data related to the NVM.

The size will be used in a future patch for implementing full NVM read
via ethtool's GEEPROM command.

The maximum possible size for the flash is bounded by the size limit for
the NVM AdminQ commands. Add a new macro, ICE_AQC_NVM_MAX_OFFSET, which
can be used to represent this upper bound.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Reviewed-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      81f07491
    • J
      ice: store NVM version info in extracted format · d4e87444
      由 Jacob Keller 提交于 
      The NVM version and Option ROM version information is stored within the
struct ice_nvm_ver_info structure. The data for the NVM is stored as
a 2byte value with the major and minor versions each using one byte from
the field. The Option ROM is stored as a 4byte value that contains
a major, build, and patch number.

Modify the code to immediately extract the version values and store them
in a new struct ice_orom_info. Remove the now unnecessary
ice_get_nvm_version function.

Update ice_ethtool.c to use the new fields directly from the structured
data.

This reduces complexity of the code that prints these versions in
ice_ethtool.c

Update the macro definitions and variable names to use the term "orom"
instead of "oem" for the Option ROM version. This helps increase the
clarity of the Option ROM version code.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Reviewed-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      d4e87444
    • J
      ice: create function to read a section of the NVM and Shadow RAM · e9450990
      由 Jacob Keller 提交于 
      The NVM contents are read via firmware by using the ice_aq_read_nvm
function. This function has a couple of limits:

1) The AdminQ commands can only take buffers sized up to 4Kb. Thus, any
   larger read must be split into multiple reads.
2) when reading from the Shadow RAM, reads must not cross sector
   boundaries. The sectors are also 4Kb in size.

Implement the ice_read_flat_nvm function to read portions of the NVM by
flat offset. That is, to read using offsets from the start of the NVM
rather than from a specific module.

This function will be able to read both from the NVM and from the Shadow
RAM. For simplicity NVM reads will always be broken up to not cross 4Kb
page boundaries, even though this is not required unless reading from
the Shadow RAM.

Use this new function as the implementation of ice_read_sr_word_aq.

The ice_read_sr_buf_aq function is not modified here. This is because
a following change will remove the only caller of that function in favor
of directly using ice_read_flat_nvm. Thus, there is little benefit to
changing it now only to remove it momentarily. At the same time, the
ice_read_sr_aq function will also be removed.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Reviewed-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      e9450990
    • J
      ice: use __le16 types for explicitly Little Endian values · 2efefb56
      由 Jacob Keller 提交于 
      The ice_read_sr_aq function returns words in the Little Endian format.
Remove the need for __force and typecasting by using a local variable in
the ice_read_sr_word_aq function.

Additionally clarify explicitly that the ice_read_sr_aq function takes
storage for __le16 values instead of using u16.

Being explicit about the endianness of this data helps when using tools
like sparse to catch endian-related issues.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Reviewed-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
      2efefb56
  15. 11 3月, 2020 7 次提交