EIP 3860: Limit and meter initcode #144
Conversation
OlivierBBB
added
hardfork shanghai
OlivierBBB
requested review from
lorenzogentile404 and
letypequividelespoubelles
|
|
For deployments: - enforce init_code_size ≤ 49152 - compute R(||init_code||) - update upfrontGasCost accordingly
OOB_INST_XCREATE no longer produces "HUB predictions".
This instruction now implicitly and silently verifies that init_code_size ≤ 49152
- new CREATE(2) maxcsx diagram - include stack/MAXCSX into case analysis - include stack/MAXCSX case in high level diagram
| \Then $\stackIcpx _{i} = 0$ | ||
| \item | ||
| \If $\stackInst _{i} \neq \inst{RETURN}$ \et $\stackCreateFlag _{i} = 0$ | ||
| \Then $\stackMaxcsx _{i} = 0$ |
There was a problem hiding this comment.
Previously only RETURN could trigger invalidCodePrefixException and maxCodeSizeException. Now CREATE(2) can trigger maxCodeSizeException, too. So split analysis.
| @@ -38,6 +39,8 @@ | |||
| \def\locMxpMtntop {\col{\mxpMod\_mtntop}} | |||
| % | |||
| \def\locTriggerOob {\col{trigger\_\oobMod}} | |||
| \def\locTriggerOobX {\col{trigger\_\oobMod\_X}} | |||
| \def\locTriggerOobU {\col{trigger\_\oobMod\_U}} | |||
There was a problem hiding this comment.
Two different OOB module triggers to keep things clean.
U stands for unexeptional
X stands for exceptional
| \item \oobDataCol{k}, for $k = 1, 2, \dots, 9$: \godGiven{} | ||
| \item | ||
| \godGiven{} | ||
| \oobDataCol{k}, for $k = 1, 2, \dots, 10$: |
There was a problem hiding this comment.
10th OOB data column
| \subsection{For $\oobInstDeployment$} \label{oob: populating: opcodes: deployment} \input{opcodes/deployment} | ||
| \subsection{For $\oobInstXcall$} \label{oob: populating: opcodes: exceptional calls} \input{opcodes/xcall} | ||
| \subsection{For $\oobInstCall$} \label{oob: populating: opcodes: call} \input{opcodes/call} | ||
| \subsection{For $\oobInstXcreate$} \label{oob: populating: opcodes: xcreate} \input{opcodes/xcreate} |
There was a problem hiding this comment.
New OOB instruction + formatting
| + & \peekingSumStandardPrefix \\ | ||
| + & \peekContext _{i + \createExceptionCallerContextRowOffset} \\ | ||
| \end{array} \right] | ||
| \] |
There was a problem hiding this comment.
We don't break the mould for exceptional CREATE's. It is inefficient (one does not need the current context row) but it simplifies everything not to differentiate.
| @@ -6,6 +6,7 @@ | |||
| \locIsCreate & \define & \decFlag{1} _{i - \createFirstStackRowOffset} \\ | |||
| \locIsCreateTwo & \define & \decFlag{2} _{i - \createFirstStackRowOffset} \\ | |||
| \locStaticx & \define & \stackStaticx _{i - \createFirstStackRowOffset} \\ | |||
| \locMaxcsx & \define & \stackMaxcsx _{i - \createFirstStackRowOffset} \\ | |||
| \item \locTriggerMmu{} | ||
| \item \locTriggerOob{} | ||
| \item \locTriggerOobX{} | ||
| \item \locTriggerOobU{} |
There was a problem hiding this comment.
trigger_OOB now split in two (trigger_OOB_X and trigger_OOB_U) as explained earlier:
- e
Xcecptional Unexceptional
There was a problem hiding this comment.
Important: definition of the two OOB module triggers.
hub/misc/oob/opcodes/xcreate.tex
Outdated
There was a problem hiding this comment.
New HUB -> OOB interface for OOB_INST_XCREATE instruction.
| @@ -22,11 +22,12 @@ | |||
| \locAbortingCondition & \define & \oobDataCol {7} _{i} & \prediction \\ | |||
| \locFailureCondition & \define & \oobDataCol {8} _{i} & \prediction \\ | |||
| \locCreatorNonce & \define & \oobDataCol {9} _{i} \\ | |||
| \locInitCodeSize & \define & \oobDataCol {10} _{i} \\ | |||
There was a problem hiding this comment.
New 10th parameter of OOB_INST_CREATE instruction
oob/opcodes/deployment.tex
Outdated
oob/opcodes/xcall.tex
Outdated
There was a problem hiding this comment.
Inclusion of XCREATE flag into shorthands.
| @@ -215,10 +219,11 @@ | |||
| \oobCtMaxCdl & \define & 0 \\ | |||
| \oobCtMaxXcall & \define & 0 \\ | |||
| \oobCtMaxCall & \define & 2 \\ | |||
| \oobCtMaxCreate & \define & 3 \\ | |||
| \oobCtMaxXcreate & \define & 0 \\ | |||
| \oobCtMaxCreate & \define & 4 \\ | |||
There was a problem hiding this comment.
1 more row due to new comparison for OOB_INST_CREATE
| \newcommand{\maxCtTypeTwo} {\redm{8}} | ||
| \newcommand{\maxCtTypeZero} {\redm{9}} | ||
| \newcommand{\maxCtTypeOne} {\redm{10}} | ||
| \newcommand{\maxCtTypeTwo} {\redm{10}} |
There was a problem hiding this comment.
Didn't carefully count, may have to figure out during implementation.
There was a problem hiding this comment.
shoukd be ok, you added only two calls so ...
| % | ||
| \input{computations/init_code_does_not_exceed_max_size} | ||
| \input{computations/init_code_pricing} | ||
| % |
There was a problem hiding this comment.
New Pricing rows
| + \!\!\! & \locIsDep & \!\!\! \cdot \!\!\! & G_{\text{txcreate}} \\ | ||
| + \!\!\! & \locIsDep & \!\!\! \cdot \!\!\! & \locInitCodeCost \\ | ||
| + \!\!\! & & & G_{\text{transaction}} \\ |
There was a problem hiding this comment.
Extra pricing cost
There was a problem hiding this comment.
Valid (deployment) transactions must enforce init_code_size ≤ 49152.
There was a problem hiding this comment.
Transactions pay an extra
oob/opcodes/xcreate.tex
Outdated
There was a problem hiding this comment.
New OOB_INST_XCREATE instruction. Enforces that init_code_size (which at this point may still be huge, whence the hi/lo parts) be > 49152 bytes long
OlivierBBB
changed the title
| \newcommand{\maxCtTypeTwo} {\redm{8}} | ||
| \newcommand{\maxCtTypeZero} {\redm{9}} | ||
| \newcommand{\maxCtTypeOne} {\redm{10}} | ||
| \newcommand{\maxCtTypeTwo} {\redm{10}} |
There was a problem hiding this comment.
shoukd be ok, you added only two calls so ...
OlivierBBB
changed the title
Implementation of #59.