Update BPF lexer (#2004)

* bpf: highlight signed type keywords

Signed type keywords such as s16 have always been supported by BPF but
recently became more frequent. They were missing in the original lexer
so this commit adds them, with an example.

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

* bpf: highlight simpler call syntax

The lexer already supports the following syntax for call instructions:

    call bpf_skb_load_bytes#1234

but a simpler syntax without the number at the end is now also possible.
This commit adds support for it, with an example.

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

* bpf: highlight all byte-swapping instructions

The original lexer was missing support for the little-endian
byte-swapping instructions le{16,32,64}. This commit adds them, together
with the new bswap{16,32,64} instructions. The later are unconditional
byte-swapping instructions, meaning they don't depend on the
architecture.

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

* bpf: fix highlight of negative hexadecimal numbers

Hexadecimal literals can also have a sign in BPF. This commit fixes this
error in the lexer and adds a corresponding example.

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

* bpf: highlight hexadecimal jump offsets

The offset of BPF jump instructions is now sometimes displayed in
hexadecimal format. This commit adds support for that, together with an
example. It also adds an example to cover the syntax where the jump has
both an offset literal and an offset label (ex. "+1 <LBB0_2>").

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

* bpf: highlight long jumps

Jump instructions with a 32-bit offset are displayed as "gotol" instead
of "goto".

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

---------

Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>

lib/rouge/lexers/bpf.rb

@@ -9,11 +9,11 @@ class BPF < RegexLexer
       tag 'bpf'
 
       TYPE_KEYWORDS = %w(
-        u8 u16 u32 u64 ll
+        u8 u16 u32 u64 s8 s16 s32 s64 ll
       ).join('|')
 
       MISC_KEYWORDS = %w(
-        be16 be32 be64 exit lock map
+        be16 be32 be64 le16 le32 le64 bswap16 bswap32 bswap64 exit lock map
       ).join('|')
 
       state :root do
@@ -32,21 +32,21 @@ class BPF < RegexLexer
         rule %r/(call)(\s+)(\d+)/i do
           groups Keyword, Text::Whitespace, Literal::Number::Integer
         end
-        rule %r/(call)(\s+)(\w+)(#)(\d+)/i do
+        rule %r/(call)(\s+)(\w+)(?:(#)(\d+))?/i do
           groups Keyword, Text::Whitespace, Name::Builtin, Punctuation, Literal::Number::Integer
         end
 
         # Unconditional jumps
-        rule %r/(goto)(\s*)(\+\d+)?(\s*)(<?\w+>?)/i do
-          groups Keyword, Text::Whitespace, Literal::Number::Integer, Text::Whitespace, Name::Label
+        rule %r/(gotol?)(\s*)([-+]0x\w+)?([-+]\d+)?(\s*)(<?\w+>?)/i do
+          groups Keyword, Text::Whitespace, Literal::Number::Hex, Literal::Number::Integer, Text::Whitespace, Name::Label
         end
 
         # Conditional jumps
-        rule %r/(if)(\s+)([rw]\d+)(\s*)([s!=<>]+)(\s*)(0x\h+|[-]?\d+)(\s*)(goto)(\s*)(\+\d+)?(\s*)(<?\w+>?)/i do
-          groups Keyword, Text::Whitespace, Name, Text::Whitespace, Operator, Text::Whitespace, Literal::Number, Text::Whitespace, Keyword, Text::Whitespace, Literal::Number::Integer, Text::Whitespace, Name::Label
+        rule %r/(if)(\s+)([rw]\d+)(\s*)([s!=<>]+)(\s*)(0x\h+|[-]?\d+)(\s*)(gotol?)(\s*)([-+]0x\w+)?([-+]\d+)?(\s*)(<?\w+>?)/i do
+          groups Keyword, Text::Whitespace, Name, Text::Whitespace, Operator, Text::Whitespace, Literal::Number, Text::Whitespace, Keyword, Text::Whitespace, Literal::Number::Hex, Literal::Number::Integer, Text::Whitespace, Name::Label
         end
-        rule %r/(if)(\s+)([rw]\d+)(\s*)([s!=<>]+)(\s*)([rw]\d+)(\s*)(goto)(\s*)(\+\d+)?(\s*)(<?\w+>?)/i do
-          groups Keyword, Text::Whitespace, Name, Text::Whitespace, Operator, Text::Whitespace, Name, Text::Whitespace, Keyword, Text::Whitespace, Literal::Number::Integer, Text::Whitespace, Name::Label
+        rule %r/(if)(\s+)([rw]\d+)(\s*)([s!=<>]+)(\s*)([rw]\d+)(\s*)(gotol?)(\s*)([-+]0x\w+)?([-+]\d+)?(\s*)(<?\w+>?)/i do
+          groups Keyword, Text::Whitespace, Name, Text::Whitespace, Operator, Text::Whitespace, Name, Text::Whitespace, Keyword, Text::Whitespace, Literal::Number::Hex, Literal::Number::Integer, Text::Whitespace, Name::Label
         end
 
         # Dereferences
@@ -70,7 +70,7 @@ class BPF < RegexLexer
         rule %r/#{MISC_KEYWORDS}/i, Keyword
 
         # Literals and global objects (maps) refered by name
-        rule %r/(0x\h+|[-]?\d+)(\s*)(ll)?/i do
+        rule %r/([-]?0x\h+|[-]?\d+)(\s*)(ll)?/i do
           groups Literal::Number, Text::Whitespace, Keyword::Type
         end
         rule %r/(\w+)(\s*)(ll)/i do

spec/visual/samples/bpf

@@ -41,6 +41,7 @@ LBB0_4:
   r6 = *(u16 *)(r1 + 8)  // BPF_LDX | BPF_H
   r7 = *(u32 *)(r2 + 16) // BPF_LDX | BPF_W
   r8 = *(u64 *)(r3 - 30) // BPF_LDX | BPF_DW
+  r1 = *(s32 *)(r6 +0)   // BPF_SMEM | BPF_LDX | BPF_W
 
 // ======== BPF_STX Class ========
   *(u8 *)(r0 + 0) = r7    // BPF_STX | BPF_B
@@ -54,6 +55,7 @@ LBB0_4:
 // ======== BPF_JMP Class ========
   goto Llabel0               // BPF_JA
   call 1                     // BPF_CALL
+  call bpf_skb_load_bytes    // BPF_CALL
   exit                       // BPF_EXIT
 
   if r0 == r1 goto Llabel0   // BPF_JEQ  | BPF_X
@@ -69,6 +71,8 @@ LBB0_4:
   if r8 s< r9 goto Llabel0   // BPF_JSLT | BPF_X
   if r9 s<= r10 goto Llabel0 // BPF_JSLE | BPF_X
 
+  if r9 < r1 goto -0x5 <Llabel0> // BPF_JLT | BPF_X
+
   if r0 == 0 goto Llabel0           // BPF_JEQ  | BPF_K
   if r3 != -1 goto Llabel0          // BPF_JNE  | BPF_K
 
@@ -82,11 +86,15 @@ LBB0_4:
   if r8 s< 0 goto Llabel0           // BPF_JSLT | BPF_K
   if r9 s<= -1 goto Llabel0         // BPF_JSLE | BPF_K
 
+  goto +1 <LBB0_2>            // BPF_JMP | BPF_JA
+  gotol +0x7fffffff <LBB0_2>  // BPF_JMP32 | BPF_JA
+
 // ======== BPF_ALU64 Class ========
   r0 += r1    // BPF_ADD  | BPF_X
   r1 -= r2    // BPF_SUB  | BPF_X
   r2 *= r3    // BPF_MUL  | BPF_X
   r3 /= r4    // BPF_DIV  | BPF_X
+  r1 s/= -0x5 // BPF_SDIV | BPF_X
 
 Llabel0 :
   r2 = -r2    // BPF_NEG
@@ -98,9 +106,11 @@ Llabel0 :
   r9 = r10    // BPF_MOV  | BPF_X
   r10 s>>= r0 // BPF_ARSH | BPF_X
 
-  r1 = be16 r1  // BPF_END  | BPF_TO_BE
-  r2 = be32 r2  // BPF_END  | BPF_TO_BE
-  r3 = be64 r3  // BPF_END  | BPF_TO_BE
+  r1 = be16 r1    // BPF_END | BPF_TO_BE
+  r2 = be32 r2    // BPF_END | BPF_TO_BE
+  r3 = be64 r3    // BPF_END | BPF_TO_BE
+  r1 = le16 r1    // BPF_END | BPF_TO_LE
+  r1 = bswap32 r1 // BPF_ALU64 | BPF_END | BPF_TO_BE
 
   r0 += 1           // BPF_ADD  | BPF_K
   r1 -= 0x1         // BPF_SUB  | BPF_K