API additions to allow enum value introspection#1
Closed
leovitch wants to merge 6 commits intothomasvl:masterfrom
Closed
API additions to allow enum value introspection#1leovitch wants to merge 6 commits intothomasvl:masterfrom
leovitch wants to merge 6 commits intothomasvl:masterfrom
Conversation
Sync to thomasvl/protobuf head
…lues. Refactored implementation so that this contains a minimum of added code. Clarified comments regarding behavior in the presence of the alias_allowed option. Added unit tests for new functionality and for the alias case.
thomasvl
reviewed
May 23, 2018
objectivec/GPBDescriptor.m
Outdated
| @@ -834,9 +834,7 @@ - (NSString *)enumNameForValue:(int32_t)number { | |||
|
|
|||
Owner
There was a problem hiding this comment.
Take out the call to calcValueNameOffsets now since this isn't directly using nameOffsets_ any more
thomasvl
reviewed
May 23, 2018
src/google/protobuf/unittest.proto
Outdated
| NEG = -1; // Intentionally negative. | ||
| } | ||
|
|
||
| enum NestedEnumAllowingAlias { |
Owner
There was a problem hiding this comment.
I try not to modify the core protos since the c++ tests drive them, looks like you can use https://github.com/google/protobuf/blob/master/src/google/protobuf/unittest.proto#L518-L526 instead.
Owner
|
Once things are fixed up, go ahead and send it over to google/protobuf, and I can land it. |
Author
|
Fixes applied, generating pull request momentarily.
…On Wed, May 23, 2018 at 11:29 PM, Thomas Van Lenten < ***@***.***> wrote:
Once things are fixed up, go ahead and send it over to google/protobuf,
and I can land it.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#1 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AAZTqTn-voo5VcSYECx_ws6ziwrIjc7oks5t1XJGgaJpZM4UJrEm>
.
|
Author
|
protocolbuffers#4678 sent.
…On Fri, May 25, 2018 at 2:07 PM, Leo Hourvitz ***@***.***> wrote:
Fixes applied, generating pull request momentarily.
On Wed, May 23, 2018 at 11:29 PM, Thomas Van Lenten <
***@***.***> wrote:
> Once things are fixed up, go ahead and send it over to google/protobuf,
> and I can land it.
>
> —
> You are receiving this because you authored the thread.
> Reply to this email directly, view it on GitHub
> <#1 (comment)>, or mute
> the thread
> <https://github.com/notifications/unsubscribe-auth/AAZTqTn-voo5VcSYECx_ws6ziwrIjc7oks5t1XJGgaJpZM4UJrEm>
> .
>
|
thomasvl
pushed a commit
that referenced
this pull request
Nov 4, 2019
thomasvl
pushed a commit
that referenced
this pull request
Jan 8, 2024
Before, every charAt would emit (on android):
```
0x00002104 adrp x17, #+0x1000 (addr 0x3000)
0x00002108 ldr w17, [x17, protocolbuffers#20]
0x0000210c ldr x0, [x0, protocolbuffers#128]
0x00002110 ldr x0, [x0, protocolbuffers#328]
0x00002114 ldr lr, [x0, protocolbuffers#24]
0x00002118 blr lr <-- Call into String.charAt(int)
```
Now, it emits the inlined implementation of charAt (branch is for possibly compressed strings):
```
0x000020b4 ldur w16, [x4, #-8]
0x000020b8 tbnz w16, #0, #+0xc (addr 0x20c4)
0x000020bc ldrb w4, [x4, x0]
0x000020c0 b #+0x8 (addr 0x20c8)
0x000020c4 ldrh w4, [x4, x0, lsl #1]
```
PiperOrigin-RevId: 591147406
thomasvl
pushed a commit
that referenced
this pull request
Aug 22, 2024
PiperOrigin-RevId: 663974332
thomasvl
pushed a commit
that referenced
this pull request
Jan 2, 2025
On a Cortex-A55 this resulted in a 28.30% reduction in CPU and wall time for the binary search path.
Loop body before:
```
.LBB0_2:
add w8, w12, #1
cmp w8, w11
b.gt .LBB0_6 // Predictable branch, ends the loop
.LBB0_3:
add w12, w8, w11
add w12, w12, w12, lsr protocolbuffers#31
asr w12, w12, #1
smaddl x0, w12, w10, x9
ldr w13, [x0]
cmp w13, w1
b.lo .LBB0_2 // Unpredictable branch here! Will be hit 50/50 in prod
b.ls .LBB0_7 // Predictable branch - ends the loop
sub w11, w12, #1
cmp w8, w11
b.le .LBB0_3 // Predictable branch - continues the loop
```
Loop body after:
```
.LBB7_1:
cmp w9, w11
b.hi .LBB7_4 // Predictable branch - ends the loop
add w12, w9, w11
lsr w12, w12, #1
umaddl x0, w12, w8, x10
sub w14, w12, #1
ldr w13, [x0]
cmp w13, w1
csel w11, w14, w11, hs
csinc w9, w9, w12, hs
b.ne .LBB7_1 // Predictable branch - continues the loop
```
PiperOrigin-RevId: 703214356
thomasvl
pushed a commit
that referenced
this pull request
Jun 11, 2025
There is no need to keep these inline as they are not specialized to the function. We also don't need branch prediction slots for them as the `ret` will always be correctly predicted, and the "end of buffer" branch is rarely taken.
With these changes, the functions to parse fixed fields are about as short and optimal as they could be:
<details>
<summary>x86-64 Assembly</summary>
```
0000000000000000 <upb_DecodeFast_Fixed32_Scalar_Tag1Byte>:
0: 45 84 c9 test r9b, r9b
3: 0f 85 00 00 00 00 jne 0x9 <upb_DecodeFast_Fixed32_Scalar_Tag1Byte+0x9>
0000000000000005: R_X86_64_PLT32 _upb_FastDecoder_DecodeGeneric-0x4
9: 4c 89 c8 mov rax, r9
c: 48 c1 e8 30 shr rax, 0x30
10: 45 89 ca mov r10d, r9d
13: 41 c1 ea 18 shr r10d, 0x18
17: 4d 0f ab d0 bts r8, r10
1b: 44 8b 56 01 mov r10d, dword ptr [rsi + 0x1]
1f: 44 89 14 02 mov dword ptr [rdx + rax], r10d
23: 48 83 c6 05 add rsi, 0x5
27: 48 3b 77 08 cmp rsi, qword ptr [rdi + 0x8]
2b: 0f 83 00 00 00 00 jae 0x31 <upb_DecodeFast_Fixed32_Scalar_Tag1Byte+0x31>
000000000000002d: R_X86_64_PLT32 upb_DecodeFast_MessageIsDoneFallback-0x4
31: 0f b7 06 movzx eax, word ptr [rsi]
34: 49 89 c9 mov r9, rcx
37: 49 c1 f9 08 sar r9, 0x8
3b: 41 89 c2 mov r10d, eax
3e: 41 21 ca and r10d, ecx
41: 41 81 e2 f8 00 00 00 and r10d, 0xf8
48: 4f 8b 5c 51 20 mov r11, qword ptr [r9 + 2*r10 + 0x20]
4d: 4f 8b 4c 51 18 mov r9, qword ptr [r9 + 2*r10 + 0x18]
52: 49 31 c1 xor r9, rax
55: 41 ff e3 jmp r11
```
</details>
<details>
<summary>ARM64 Assembly</summary>
```
0000000000000000 <upb_DecodeFast_Fixed32_Scalar_Tag1Byte>:
0: f2401cbf tst x5, #0xff
4: 54000040 b.eq 0xc <upb_DecodeFast_Fixed32_Scalar_Tag1Byte+0xc>
8: 14000000 b 0x8 <upb_DecodeFast_Fixed32_Scalar_Tag1Byte+0x8>
0000000000000008: R_AARCH64_JUMP26 _upb_FastDecoder_DecodeGeneric
c: d370fca8 lsr x8, x5, protocolbuffers#48
10: b840102 ldur w9, [x1, #0x1]
14: 5280002a mov w10, #0x1 // =1
18: 91001421 add x1, x1, #0x5
1c: b8286849 str w9, [x2, x8]
20: d358fca8 lsr x8, x5, protocolbuffers#24
24: f9400409 ldr x9, [x0, #0x8]
28: 9ac82148 lsl x8, x10, x8
2c: eb01013f cmp x9, x1
30: 54000149 b.ls 0x58 <upb_DecodeFast_Fixed32_Scalar_Tag1Byte+0x58>
34: 79400029 ldrh w9, [x1]
38: 9348fc6a asr x10, x3, protocolbuffers#8
3c: 12001c6b and w11, w3, #0xff
40: aa040104 orr x4, x8, x4
44: 8a09016b and x11, x11, x9
48: 8b0b054a add x10, x10, x11, lsl #1
4c: a941994b ldp x11, x6, [x10, #0x18]
50: ca090165 eor x5, x11, x9
54: d61f00c0 br x6
58: aa040104 orr x4, x8, x4
5c: 14000000 b 0x5c <upb_DecodeFast_Fixed32_Scalar_Tag1Byte+0x5c>
000000000000005c: R_AARCH64_JUMP26 upb_DecodeFast_MessageIsDoneFallback
```
</details>
PiperOrigin-RevId: 765632464
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Here's a proposal for a minimal API to allow introspecting the list of enum values. Let me know if you have suggestions.