Skip to content
GitLab
Explore
Sign in
Primary navigation
Search or go to…
Project
I
iGPS
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Model registry
Operate
Environments
Monitor
Incidents
Service Desk
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Terms and privacy
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
FP
iGPS
Commits
ea9a85e0
Commit
ea9a85e0
authored
9 years ago
by
Janno
Browse files
Options
Downloads
Patches
Plain Diff
GFork proof, WIP
parent
cd92b558
No related branches found
Branches containing commit
No related tags found
Tags containing commit
No related merge requests found
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
framework.tex
+16
-10
16 additions, 10 deletions
framework.tex
layer1def.tex
+10
-2
10 additions, 2 deletions
layer1def.tex
with
26 additions
and
12 deletions
framework.tex
+
16
−
10
View file @
ea9a85e0
...
@@ -252,8 +252,7 @@ We split the construction of ghost space into two layers.
...
@@ -252,8 +252,7 @@ We split the construction of ghost space into two layers.
%
%
\begin{enumerate}
\begin{enumerate}
\item
\item
We introduce lower bounds for buffers so that the buffers can be updated concurrently by
We introduce lower bounds for buffers so that the buffers can be updated concurrently.
processors and the communication process.
At the same time, we introduce logical per-location histories.
At the same time, we introduce logical per-location histories.
In these, we keep track of the timestamps, values, and the committing processor's buffer (including the new write event).
In these, we keep track of the timestamps, values, and the committing processor's buffer (including the new write event).
We call the entries of these histories historic events.
We call the entries of these histories historic events.
...
@@ -353,7 +352,7 @@ every write event in a processor's buffer has to be justified by a write event i
...
@@ -353,7 +352,7 @@ every write event in a processor's buffer has to be justified by a write event i
\DBHInv
(
\fMB
,
\fH
)
\eqdef
{}&
\All
{{
\proc
\in
\dom
{
\fMB
}}}
.
\fMB
(
\proc
)
\lesssim
\fH
\DBHInv
(
\fMB
,
\fH
)
\eqdef
{}&
\All
{{
\proc
\in
\dom
{
\fMB
}}}
.
\fMB
(
\proc
)
\lesssim
\fH
\end
{
array
}\]
\end
{
array
}\]
%
%
\paragraph
{
Intution
}
The
$
\lesssim
$
relation enforces that, if processor
$
\pi
$
commits a write event, every processor that learns of this write event
\paragraph
{
Intu
i
tion
}
The
$
\lesssim
$
relation enforces that, if processor
$
\pi
$
commits a write event, every processor that learns of this write event
must also have at least processor
$
\pi
$
's knowledge (in terms of write events) up to that event.
must also have at least processor
$
\pi
$
's knowledge (in terms of write events) up to that event.
This knowledge is represented by the buffer in the location's history.
This knowledge is represented by the buffer in the location's history.
...
@@ -399,10 +398,9 @@ reduces to $\unitval$.
...
@@ -399,10 +398,9 @@ reduces to $\unitval$.
%
%
%
%
\subsection
{
Triples
}
\subsection
{
Triples
}
We prove our first set of triples.
We provide the first set of triples.
This layer's triples are not sufficiently high-level enough to express their pre- and post-conditions purely in GPS assertions.
The pre- and post-conditions of the triples are not monotone buffer predicates.
Thus, we make use of the injection of predicates over buffers defined in Section~
\ref
{
sec:new-assertions
}
.
Thus, the triples are
\textbf
{
not
}
representable in GPS.
We skip the binder of the predicate whenever the meta-level assertions does not depend on the buffer.
\begin{mathpar}
\begin{mathpar}
\inferH
{
AtomicRead2
}{}{
\ereadHyp
\vdash
\hoare
{
\ereadPreB
\inferH
{
AtomicRead2
}{}{
\ereadHyp
\vdash
\hoare
{
\ereadPreB
...
@@ -411,8 +409,10 @@ We skip the binder of the predicate whenever the meta-level assertions does not
...
@@ -411,8 +409,10 @@ We skip the binder of the predicate whenever the meta-level assertions does not
}{
\estore
{
\wtat
}{
\loc
}{
\val
}
,
\proc
}{
\estorePostB
}}
\\
}{
\estore
{
\wtat
}{
\loc
}{
\val
}
,
\proc
}{
\estorePostB
}}
\\
\and\inferH
{
CAS2
}{}{
\ecasHyp
\vdash
\hoare
{
\ecasPreB
}{
\escas
{
\loc
}{
\val
_
o
}{
\val
_
n
}}{
\ecasPostB
}}
\\
\and\inferH
{
CAS2
}{}{
\ecasHyp
\vdash
\hoare
{
\ecasPreB
}{
\escas
{
\loc
}{
\val
_
o
}{
\val
_
n
}}{
\ecasPostB
}}
\\
\and\inferH
{
Process2
}{}{
\processHyp
\vdash
\hoare
{
\top
}{
\eprocess
,
\proc
}{
\processPostB
}}
\and\inferH
{
Process2
}{}{
\processHyp
\vdash
\hoare
{
\top
}{
\eprocess
,
\proc
}{
\processPostB
}}
\and\inferH
{
Fork2
}{
\forkUHyp
\vdash
\hoare
{
\forkUPreB
}{
e,
\procB
}{
\forkUPostB
}}
\and\inferH
{
GFork2
}{
\gforkUHyp
\vdash
\hoare
{
\gforkUPreB
}{
e,
\procB
}{
\gforkUPostB
}}
{
\forkHyp
\vdash
\hoare
{
\forkPreB
}{
\gfork
(e),
\proc
}{
\forkPostB
}}
\\
{
\gforkHyp
\vdash
\hoare
{
\gforkPreB
}{
\gfork
(e),
\proc
}{
\gforkPostB
}}
\\
\and\inferH
{
Fork2
}{
\forkUHyp
\vdash
\hoare
{
\forkUPreB
}{
e,
\proc
}{
\forkUPostB
}}
{
\forkHyp
\vdash
\hoare
{
\forkPreB
}{
\fork
(e),
\proc
}{
\forkPostB
}}
\\
\end{mathpar}
\end{mathpar}
...
@@ -840,7 +840,13 @@ Trivial. \janno{TODO}
...
@@ -840,7 +840,13 @@ Trivial. \janno{TODO}
\pline
[\top]
{
\top
}
\pline
[\top]
{
\top
}
\end{hproofenv}
\end{hproofenv}
%
%
%
\subsubsection
{
GFork
}
\begin{hproofenv}
Context:
$
\buf
$
,
$
\proc
$
,
$
\bseen
{
\proc
}{
\buf
}$
~
\\
Context:
$
\All
\procB
.
\gforkU
$
~
\\
\
\end{hproofenv}
%
%
%
\clearpage
\clearpage
\newcommand
{
\Prot
}
[0]
{
\mathrm
{
PROT
}}
\newcommand
{
\Prot
}
[0]
{
\mathrm
{
PROT
}}
...
...
This diff is collapsed.
Click to expand it.
layer1def.tex
+
10
−
2
View file @
ea9a85e0
...
@@ -57,12 +57,20 @@
...
@@ -57,12 +57,20 @@
\lor
&
\;
(
\val
' =
\None
\ast
\hist
' =
\hist
)
\lor
&
\;
(
\val
' =
\None
\ast
\hist
' =
\hist
)
\end{aligned}
\end{aligned}
}
}
\newcommand\forkUHyp
{
\bseen
{
\procB
}{
\buf
'
}}
\newcommand\gforkUHyp
{
\bseen
{
\procB
}{
\buf
'
}}
\newcommand\gforkUPreB
{
Q
}
\newcommand\gforkUPostB
{
\top
}
\newcommand\gforkU
[0]
{
\gforkUHyp
\vdash
\hoare
{
\gforkUPreB
}{
\expr
,
\procB
}{
\gforkUPostB
}}
\newcommand\gforkHyp
{
\bseen
{
\proc
}{
\buf
}}
\newcommand\gforkPreB
{
Q
}
\newcommand\gforkPostB
{
\Ret\rho
.
\Exists
\buf
'
\bsup
\buf
.
\bseen
{
\proc
}{
\buf
'
}
\ast
\bseen
{
\procB
}{
\buf
'
}}
\newcommand\forkUHyp
{
\bseen
{
\proc
}{
\buf
'
}}
\newcommand\forkUPreB
{
Q
}
\newcommand\forkUPreB
{
Q
}
\newcommand\forkUPostB
{
\top
}
\newcommand\forkUPostB
{
\top
}
\newcommand\forkHyp
{
\bseen
{
\proc
}{
\buf
}}
\newcommand\forkHyp
{
\bseen
{
\proc
}{
\buf
}}
\newcommand\forkPreB
{
Q
}
\newcommand\forkPreB
{
Q
}
\newcommand\forkPostB
{
\Ret\
rho
.
\Exists
\buf
'
\bsup
\buf
.
\bseen
{
\proc
}{
\buf
'
}
\ast
\bseen
{
\procB
}{
\buf
'
}
}
\newcommand\forkPostB
{
\Ret\
unitval
.
\Exists
\buf
'
\bsup
\buf
.
\bseen
{
\proc
}{
\buf
'
}}
\newcommand\processHyp
{
\bseen
{
\proc
}{
\buf
}}
\newcommand\processHyp
{
\bseen
{
\proc
}{
\buf
}}
\newcommand\processPreB
{
\bseen
{
\proc
}{
\buf
}}
\newcommand\processPreB
{
\bseen
{
\proc
}{
\buf
}}
...
...
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment