This fixues up the markup to fix a warning, be more consistent with use of monospace, and use the correct .rst syntax for <em> (* instead of _). It also clarifies the explanation of Clang's -mcpu requirements for this feature, Alexei pointed out that use of the word "version" was confusing here. NB this conflicts with Lukas' patch at [1], here where I've added `::` to fix the warning, I also kept the original ':' which appears in the output text. [1] https://lore.kernel.org/bpf/CA+i-1C3cEXqxcXfD4sibQfx+dtmmzvOzruhk8J5pAw3g5v=KgA@xxxxxxxxxxxxxx/T/#t Signed-off-by: Brendan Jackman <jackmanb@xxxxxxxxxx> --- Documentation/networking/filter.rst | 30 +++++++++++++++-------------- 1 file changed, 16 insertions(+), 14 deletions(-) diff --git a/Documentation/networking/filter.rst b/Documentation/networking/filter.rst index f6d8f90e9a56..ba03e90a9163 100644 --- a/Documentation/networking/filter.rst +++ b/Documentation/networking/filter.rst @@ -1048,12 +1048,12 @@ Unlike classic BPF instruction set, eBPF has generic load/store operations:: Where size is one of: BPF_B or BPF_H or BPF_W or BPF_DW. It also includes atomic operations, which use the immediate field for extra -encoding. +encoding: :: .imm = BPF_ADD, .code = BPF_ATOMIC | BPF_W | BPF_STX: lock xadd *(u32 *)(dst_reg + off16) += src_reg .imm = BPF_ADD, .code = BPF_ATOMIC | BPF_DW | BPF_STX: lock xadd *(u64 *)(dst_reg + off16) += src_reg -The basic atomic operations supported are: +The basic atomic operations supported are: :: BPF_ADD BPF_AND @@ -1066,12 +1066,12 @@ memory location addresed by ``dst_reg + off`` is atomically modified, with immediate, then these operations also overwrite ``src_reg`` with the value that was in memory before it was modified. -The more special operations are: +The more special operations are: :: BPF_XCHG This atomically exchanges ``src_reg`` with the value addressed by ``dst_reg + -off``. +off``. :: BPF_CMPXCHG @@ -1081,19 +1081,21 @@ before is loaded back to ``R0``. Note that 1 and 2 byte atomic operations are not supported. -Except ``BPF_ADD`` _without_ ``BPF_FETCH`` (for legacy reasons), all 4 byte -atomic operations require alu32 mode. Clang enables this mode by default in -architecture v3 (``-mcpu=v3``). For older versions it can be enabled with -``-Xclang -target-feature -Xclang +alu32``. +Clang can generate atomic instructions when ``-mcpu=v3`` is enabled (this is the +default). If a lower version for ``-mcpu`` is set, the only atomic instruction +Clang can generate is ``BPF_ADD`` *without* ``BPF_FETCH``. If you need to +enable the atomics features, while keeping a lower ``-mcpu`` version, you can +use ``-Xclang -target-feature -Xclang +alu32``. -You may encounter BPF_XADD - this is a legacy name for BPF_ATOMIC, referring to -the exclusive-add operation encoded when the immediate field is zero. +You may encounter ``BPF_XADD`` - this is a legacy name for ``BPF_ATOMIC``, +referring to the exclusive-add operation encoded when the immediate field is +zero. -eBPF has one 16-byte instruction: BPF_LD | BPF_DW | BPF_IMM which consists +eBPF has one 16-byte instruction: ``BPF_LD | BPF_DW | BPF_IMM`` which consists of two consecutive ``struct bpf_insn`` 8-byte blocks and interpreted as single -instruction that loads 64-bit immediate value into a dst_reg. -Classic BPF has similar instruction: BPF_LD | BPF_W | BPF_IMM which loads -32-bit immediate value into a register. +instruction that loads 64-bit immediate value into a dst_reg. Classic BPF has +similar instruction: ``BPF_LD | BPF_W | BPF_IMM`` which loads 32-bit immediate +value into a register. eBPF verifier ------------- base-commit: 232164e041e925a920bfd28e63d5233cfad90b73 -- 2.30.0.284.gd98b1dd5eaa7-goog
