- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
doc/theses/thierry_delisle_PhD/thesis/text/io.tex
rd895e32 ra44514e 1 1 \chapter{User Level \io}\label{userio} 2 2 As mentioned in Section~\ref{prev:io}, user-level \io requires multiplexing the \io operations of many \ats onto fewer \glspl{proc} using asynchronous \io operations. 3 Different operating systems offer various forms of asynchronous operations and, as mentioned in Chapter~\ref{intro}, this work is exclusively focused on the Linux operating 3 Different operating systems offer various forms of asynchronous operations and, as mentioned in Chapter~\ref{intro}, this work is exclusively focused on the Linux operating-system. 4 4 5 5 \section{Kernel Interface} … … 13 13 In this context, ready means \emph{some} operation can be performed without blocking. 14 14 It does not mean an operation returning \lstinline{EAGAIN} succeeds on the next try. 15 For example, a ready read may only return a subset of requested bytes and the read must be issue dagain for the remaining bytes, at which point it may return \lstinline{EAGAIN}.}15 For example, a ready read may only return a subset of requested bytes and the read must be issues again for the remaining bytes, at which point it may return \lstinline{EAGAIN}.} 16 16 This mechanism is also crucial in determining when all \ats are blocked and the application \glspl{kthrd} can now block. 17 17 18 18 There are three options to monitor file descriptors in Linux:\footnote{ 19 19 For simplicity, this section omits \lstinline{pselect} and \lstinline{ppoll}. 20 The difference between these system calls and \lstinline{select} and \lstinline{poll}, respectively, is not relevant for this discussion.} 20 The difference between these system calls and \lstinline{select} and \lstinline{poll}, respectively, is not relevant for this discussion.}, 21 21 @select@~\cite{MAN:select}, @poll@~\cite{MAN:poll} and @epoll@~\cite{MAN:epoll}. 22 22 All three of these options offer a system call that blocks a \gls{kthrd} until at least one of many file descriptors becomes ready. … … 33 33 Often the I/O manager has a timeout, polls, or is sent a signal on changes to mitigate this problem. 34 34 35 % \begin{comment} 36 % From: Tim Brecht <brecht@uwaterloo.ca> 37 % Subject: Re: FD sets 38 % Date: Wed, 6 Jul 2022 00:29:41 +0000 39 40 % Large number of open files 41 % -------------------------- 42 43 % In order to be able to use more than the default number of open file 44 % descriptors you may need to: 45 46 % o increase the limit on the total number of open files /proc/sys/fs/file-max 47 % (on Linux systems) 48 49 % o increase the size of FD_SETSIZE 50 % - the way I often do this is to figure out which include file __FD_SETSIZE 51 % is defined in, copy that file into an appropriate directory in ./include, 52 % and then modify it so that if you use -DBIGGER_FD_SETSIZE the larger size 53 % gets used 54 55 % For example on a RH 9.0 distribution I've copied 56 % /usr/include/bits/typesizes.h into ./include/i386-linux/bits/typesizes.h 57 58 % Then I modify typesizes.h to look something like: 59 60 % #ifdef BIGGER_FD_SETSIZE 61 % #define __FD_SETSIZE 32767 62 % #else 63 % #define __FD_SETSIZE 1024 64 % #endif 65 66 % Note that the since I'm moving and testing the userver on may different 67 % machines the Makefiles are set up to use -I ./include/$(HOSTTYPE) 68 69 % This way if you redefine the FD_SETSIZE it will get used instead of the 70 % default original file. 71 % \end{comment} 72 35 73 \paragraph{\lstinline{poll}} is the next oldest option, and takes as input an array of structures containing the FD numbers rather than their position in an array of bits, allowing a more compact input for interest sets that contain widely spaced FDs. 36 74 For small interest sets with densely packed FDs, the @select@ bit mask can take less storage, and hence, copy less information into the kernel. 37 Furthermore, @poll@ is non-destructive, so the array of structures does not have to be re-initialize don every call.38 Like @select@, @poll@ suffers from the limitation that the interest set cannot be changed by other \gls pl{kthrd}, while a manager thread is blocked in @poll@.75 Furthermore, @poll@ is non-destructive, so the array of structures does not have to be re-initialize on every call. 76 Like @select@, @poll@ suffers from the limitation that the interest set cannot be changed by other \gls{kthrd}, while a manager thread is blocked in @poll@. 39 77 40 78 \paragraph{\lstinline{epoll}} follows after @poll@, and places the interest set in the kernel rather than the application, where it is managed by an internal \gls{kthrd}. … … 52 90 An alternative to @O_NONBLOCK@ is the AIO interface. 53 91 Its interface lets programmers enqueue operations to be performed asynchronously by the kernel. 54 Completions of these operations can be communicated in various ways: either by spawning a new \gls{kthrd}, sending a Linux signal, or polling for completion of one or more operations.92 Completions of these operations can be communicated in various ways: either by spawning a new \gls{kthrd}, sending a Linux signal, or by polling for completion of one or more operation. 55 93 For this work, spawning a new \gls{kthrd} is counter-productive but a related solution is discussed in Section~\ref{io:morethreads}. 56 Using interrupt handlers can also lead to fairly complicated interactions between subsystems and has anon-trivial cost.94 Using interrupts handlers can also lead to fairly complicated interactions between subsystems and has non-trivial cost. 57 95 Leaving polling for completion, which is similar to the previous system calls. 58 96 AIO only supports read and write operations to file descriptors, it does not have the same limitation as @O_NONBLOCK@, \ie, the file descriptors can be regular files and blocked devices. 59 97 It also supports batching multiple operations in a single system call. 60 98 61 AIO offers two different approaches to polling: @aio_error@ can be used as a spinning form of polling, returning @EINPROGRESS@ until the operation is completed, and @aio_suspend@ can be used similarly to @select@, @poll@ or @epoll@, to wait until one or more requests have beencompleted.62 For \io multiplexing, @aio_suspend@ is the best interface.99 AIO offers two different approaches to polling: @aio_error@ can be used as a spinning form of polling, returning @EINPROGRESS@ until the operation is completed, and @aio_suspend@ can be used similarly to @select@, @poll@ or @epoll@, to wait until one or more requests have completed. 100 For the purpose of \io multiplexing, @aio_suspend@ is the best interface. 63 101 However, even if AIO requests can be submitted concurrently, @aio_suspend@ suffers from the same limitation as @select@ and @poll@, \ie, the interest set cannot be dynamically changed while a call to @aio_suspend@ is in progress. 64 AIO also suffers from the limitation of specifying which requests have beencompleted, \ie programmers have to poll each request in the interest set using @aio_error@ to identify the completed requests.102 AIO also suffers from the limitation of specifying which requests have completed, \ie programmers have to poll each request in the interest set using @aio_error@ to identify the completed requests. 65 103 This limitation means that, like @select@ and @poll@ but not @epoll@, the time needed to examine polling results increases based on the total number of requests monitored, not the number of completed requests. 66 104 Finally, AIO does not seem to be a popular interface, which I believe is due in part to this poor polling interface. … … 86 124 in 87 125 ``some kind of arbitrary \textit{queue up asynchronous system call} model''. 88 This description is quite close to the interface described in the next section.126 This description is actually quite close to the interface described in the next section. 89 127 90 128 \subsection{\lstinline{io_uring}} … … 97 135 In addition to supporting reads and writes to any file descriptor like AIO, it supports other operations like @open@, @close@, @fsync@, @accept@, @connect@, @send@, @recv@, @splice@, \etc. 98 136 99 On top of these, @io_uring@ adds many extras like avoiding copies between the kernel and user space using shared memory, allowing different mechanisms to communicate with device drivers, and supporting chains of requests, \ie, requests that automatically trigger follow-up requests on completion.137 On top of these, @io_uring@ adds many extras like avoiding copies between the kernel and user-space using shared memory, allowing different mechanisms to communicate with device drivers, and supporting chains of requests, \ie, requests that automatically trigger followup requests on completion. 100 138 101 139 \subsection{Extra Kernel Threads}\label{io:morethreads} … … 105 143 This approach is used by languages like Go~\cite{GITHUB:go}, frameworks like libuv~\cite{libuv}, and web servers like Apache~\cite{apache} and NGINX~\cite{nginx}, since it has the advantage that it can easily be used across multiple operating systems. 106 144 This advantage is especially relevant for languages like Go, which offer a homogeneous \glsxtrshort{api} across all platforms. 107 As opposed to C, which has a very limited standard \glsxtrshort{api}for \io, \eg, the C standard library has no networking.145 As opposed to C, which has a very limited standard api for \io, \eg, the C standard library has no networking. 108 146 109 147 \subsection{Discussion} … … 118 156 An event engine's responsibility is to use the kernel interface to multiplex many \io operations onto few \glspl{kthrd}. 119 157 In concrete terms, this means \ats enter the engine through an interface, the event engine then starts an operation and parks the calling \ats, returning control to the \gls{proc}. 120 The parked \ats are then rescheduled by the event engine once the desired operation has beencompleted.158 The parked \ats are then rescheduled by the event engine once the desired operation has completed. 121 159 122 160 \subsection{\lstinline{io_uring} in depth}\label{iouring} … … 133 171 \centering 134 172 \input{io_uring.pstex_t} 135 \caption[Overview of \lstinline{io_uring}]{Overview of \lstinline{io_uring} \smallskip\newline Two ring buffer sare used to communicate with the kernel, one for completions~(right) and one for submissions~(left). The submission ring indexes into a pre-allocated array (denoted \emph{S}) instead.}173 \caption[Overview of \lstinline{io_uring}]{Overview of \lstinline{io_uring} \smallskip\newline Two ring buffer are used to communicate with the kernel, one for completions~(right) and one for submissions~(left). The submission ring indexes into a pre-allocated array (denoted \emph{S}) instead.} 136 174 \label{fig:iouring} 137 175 \end{figure} … … 146 184 \item 147 185 The SQE is filled according to the desired operation. 148 This step is straight forward.149 The only detail worth mentioning is that SQEs have a @user_data@ field that must be filled to match submission and completion entries.186 This step is straight forward. 187 The only detail worth mentioning is that SQEs have a @user_data@ field that must be filled in order to match submission and completion entries. 150 188 \item 151 189 The SQE is submitted to the submission ring by appending the index of the SQE to the ring following regular ring buffer steps: \lstinline{buffer[head] = item; head++}. … … 169 207 170 208 The @io_uring_enter@ system call is protected by a lock inside the kernel. 171 This protection means that concurrent call sto @io_uring_enter@ using the same instance are possible, but there is no performance gained from parallel calls to @io_uring_enter@.209 This protection means that concurrent call to @io_uring_enter@ using the same instance are possible, but there is no performance gained from parallel calls to @io_uring_enter@. 172 210 It is possible to do the first three submission steps in parallel; 173 211 however, doing so requires careful synchronization. … … 178 216 This restriction means \io request bursts may have to be subdivided and submitted in chunks at a later time. 179 217 180 An important detail to keep in mind is that just like ``The cloud is just someone else's computer''\cite{xkcd:cloud}, asynchronous operations are just operation susing someone else's threads.181 Indeed, asynchronous operation scan require computation time to complete, which means that if this time is not taken from the thread that triggered the asynchronous operation, it must be taken from some other threads.218 An important detail to keep in mind is that just like ``The cloud is just someone else's computer''\cite{xkcd:cloud}, asynchronous operations are just operation using someone else's threads. 219 Indeed, asynchronous operation can require computation time to complete, which means that if this time is not taken from the thread that triggered the asynchronous operation, it must be taken from some other threads. 182 220 In this case, the @io_uring@ operations that cannot be handled directly in the system call must be delegated to some other \gls{kthrd}. 183 221 To this end, @io_uring@ maintains multiple \glspl{kthrd} inside the kernel that are not exposed to the user. 184 Th ree kindsof operations that can need the \glspl{kthrd}:222 There are three kind of operations that can need the \glspl{kthrd}: 185 223 186 224 \paragraph{Operations using} @IOSQE_ASYNC@. … … 190 228 This is also a fairly simple case. As mentioned earlier in this chapter, [@O_NONBLOCK@] has no effect for regular files and block devices. 191 229 @io_uring@ must also take this reality into account by delegating operations on regular files and block devices. 192 In fact ,@io_uring@ maintains a pool of \glspl{kthrd} dedicated to these operations, which are referred to as \newterm{bounded workers}.230 In fact @io_uring@ maintains a pool of \glspl{kthrd} dedicated to these operations, which are referred to as \newterm{bounded workers}. 193 231 194 232 \paragraph{Unbounded operations that must be retried.} … … 197 235 @io_uring@ maintains a separate pool for these operations. 198 236 The \glspl{kthrd} in this pool are referred to as \newterm{unbounded workers}. 199 Unbounded workers are also responsible forhandling operations using @IOSQE_ASYNC@.237 Unbounded workers are also responsible of handling operations using @IOSQE_ASYNC@. 200 238 201 239 @io_uring@ implicitly spawns and joins both the bounded and unbounded workers based on its evaluation of the needs of the workload. 202 240 This effectively encapsulates the work that is needed when using @epoll@. 203 Indeed, @io_uring@ does not change Linux's underlying handling of \io ope rations, it simply offers an asynchronous \glsxtrshort{api} on top of the existing system.241 Indeed, @io_uring@ does not change Linux's underlying handling of \io opeartions, it simply offers an asynchronous \glsxtrshort{api} on top of the existing system. 204 242 205 243 206 244 \subsection{Multiplexing \io: Submission} 207 245 208 The submission side is the most complicated aspect of @io_uring@ and the completion side effectively follows from the design decisions made on the submission side.246 The submission side is the most complicated aspect of @io_uring@ and the completion side effectively follows from the design decisions made in the submission side. 209 247 While there is freedom in designing the submission side, there are some realities of @io_uring@ that must be taken into account. 210 248 It is possible to do the first steps of submission in parallel; … … 217 255 As described in Chapter~\ref{practice}, this does not translate into constant CPU usage.}. 218 256 Note that once an operation completes, there is nothing that ties it to the @io_uring@ instance that handled it. 219 Nothing preventing a new operation, with for example the same file descriptor, to usea different @io_uring@ instance.257 There is nothing preventing a new operation with, \eg the same file descriptors to a different @io_uring@ instance. 220 258 221 259 A complicating aspect of submission is @io_uring@'s support for chains of operations, where the completion of an operation triggers the submission of the next operation on the link. … … 225 263 Support for this feature can be fulfilled simply by supporting arbitrary user code between allocation and submission. 226 264 227 Similar to scheduling, sharding @io_uring@ instances can be done privately, \ie, one instance per \ proc, in decoupled pools, \ie, a pool of \procs usinga pool of @io_uring@ instances without one-to-one coupling between any given instance and any given \gls{proc}, or some mix of the two.265 Similar to scheduling, sharding @io_uring@ instances can be done privately, \ie, one instance per \glspl{proc}, in decoupled pools, \ie, a pool of \glspl{proc} use a pool of @io_uring@ instances without one-to-one coupling between any given instance and any given \gls{proc}, or some mix of the two. 228 266 These three sharding approaches are analyzed. 229 267 … … 232 270 This alleviates the need for synchronization on the submissions, requiring only that \ats are not time-sliced during submission steps. 233 271 This requirement is the same as accessing @thread_local@ variables, where a \at is accessing kernel-thread data, is time-sliced, and continues execution on another kernel thread but is now accessing the wrong data. 234 This failure is the \newterm{serially reusable problem}~\cite{SeriallyReusable}.235 Hence, allocated SQEs must be submitted to the same ring on the same \gls{proc}, which effectively forces the application to submit SQEs in order of allocation.\footnote{272 This failure is the serially reusable problem~\cite{SeriallyReusable}. 273 Hence, allocated SQEs must be submitted to the same ring on the same \gls{proc}, which effectively forces the application to submit SQEs in allocation order.\footnote{ 236 274 To remove this requirement, a \at needs the ability to ``yield to a specific \gls{proc}'', \ie, \park with the guarantee it unparks on a specific \gls{proc}, \ie the \gls{proc} attached to the correct ring.} 237 275 From the subsystem's point of view, the allocation and submission are sequential, greatly simplifying both. 238 276 In this design, allocation and submission form a partitioned ring buffer as shown in Figure~\ref{fig:pring}. 239 Once added to the ring buffer, the attached \gls{proc} has a significant amount of flexibility with regard to when to perform the system call.277 Once added to the ring buffer, the attached \gls{proc} has a significant amount of flexibility with regards to when to perform the system call. 240 278 Possible options are: when the \gls{proc} runs out of \ats to run, after running a given number of \ats, \etc. 241 279 … … 243 281 \centering 244 282 \input{pivot_ring.pstex_t} 245 \caption[Partitioned ring buffer]{Partitioned ring buffer \smallskip\newline Allocated sqes are append edto the first partition.283 \caption[Partitioned ring buffer]{Partitioned ring buffer \smallskip\newline Allocated sqes are appending to the first partition. 246 284 When submitting, the partition is advanced. 247 285 The kernel considers the partition as the head of the ring.} … … 256 294 A more involved version of this approach tries to solve these problems using a pattern called \newterm{helping}. 257 295 \ats that cannot submit \io operations, either because of an allocation failure or \glslink{atmig}{migration} to a different \gls{proc} between allocation and submission, create an \io object and add it to a list of pending submissions per \gls{proc} and a list of pending allocations, probably per cluster. 258 While there is still astrong coupling between \glspl{proc} and @io_uring@ instances, these data structures allow moving \ats to a specific \gls{proc}, when the current \gls{proc} cannot fulfill the \io request.296 While there is still the strong coupling between \glspl{proc} and @io_uring@ instances, these data structures allow moving \ats to a specific \gls{proc}, when the current \gls{proc} cannot fulfill the \io request. 259 297 260 298 Imagine a simple scenario with two \ats on two \glspl{proc}, where one \at submits an \io operation and then sets a flag, while the other \at spins until the flag is set. … … 263 301 No other \gls{proc} can help the \at since @io_uring@ instances are strongly coupled to \glspl{proc}. 264 302 However, the \io \gls{proc} is unable to help because it is executing the spinning \at resulting in a deadlock. 265 While this example is artificial, in the presence of many \ats, this problem canarise ``in the wild''.303 While this example is artificial, in the presence of many \ats, it is possible for this problem to arise ``in the wild''. 266 304 Furthermore, this pattern is difficult to reliably detect and avoid. 267 Once in this situation, the only escape is to interrupt the spinning \at, either directly or via some regular preemption, \eg time slicing.305 Once in this situation, the only escape is to interrupted the spinning \at, either directly or via some regular preemption, \eg time slicing. 268 306 Having to interrupt \ats for this purpose is costly, the latency can be large between interrupts, and the situation may be hard to detect. 269 307 Interrupts are needed here entirely because the \gls{proc} is tied to an instance it is not using. … … 281 319 \item 282 320 The scheme to route \io requests to specific @io_uring@ instances does not introduce contention. 283 This aspect has oversized importance because it comes into play before the sharding of instances, and as such, all \glspl{hthrd} can contend on the routing algorithm.321 This aspect has an oversized importance because it comes into play before the sharding of instances, and as such, all \glspl{hthrd} can contend on the routing algorithm. 284 322 \end{itemize} 285 323 286 324 Allocation in this scheme is fairly easy. 287 Free SQEs, \ie, SQEs that are not currently being used to represent a request, can be written -to safely and have a field called @user_data@ that the kernel only reads to copy to CQEs.288 Allocation also does not require ordering guaranteesas all free SQEs are interchangeable.325 Free SQEs, \ie, SQEs that are not currently being used to represent a request, can be written to safely and have a field called @user_data@ that the kernel only reads to copy to @cqe@s. 326 Allocation also requires no ordering guarantee as all free SQEs are interchangeable. 289 327 The only added complexity is that the number of SQEs is fixed, which means allocation can fail. 290 328 … … 292 330 Furthermore, the routing algorithm should block operations up-front, if none of the instances have available SQEs. 293 331 294 Once an SQE is allocated, \ats insert the \io request information and keep track of the SQE index and the instance it belongs to.332 Once an SQE is allocated, \ats insert the \io request information, and keep track of the SQE index and the instance it belongs to. 295 333 296 334 Once an SQE is filled in, it is added to the submission ring buffer, an operation that is not thread-safe, and then the kernel must be notified using the @io_uring_enter@ system call. 297 The submission ring buffer is the same size as the pre-allocated SQE buffer, therefore pushing to the ring buffer cannot fail because it would mean a n SQEmultiple times in the ring buffer, which is undefined behaviour.298 However, as mentioned, the system call itself can fail with the expectation that it can be retried once some submitted operations arecomplete.335 The submission ring buffer is the same size as the pre-allocated SQE buffer, therefore pushing to the ring buffer cannot fail because it would mean a \lstinline{sqe} multiple times in the ring buffer, which is undefined behaviour. 336 However, as mentioned, the system call itself can fail with the expectation that it can be retried once some submitted operations complete. 299 337 300 338 Since multiple SQEs can be submitted to the kernel at once, it is important to strike a balance between batching and latency. 301 Operations that are ready to be submitted should be batched together in few system calls, but at the same time, operations should not be left pending for long period s before being submitted.302 Balancing submission can be handled by either designating one of the submitting \ats as the \at responsible for the system call for the current batch of SQEs or by having some other party regularly submitall ready SQEs, \eg, the poller \at mentioned later in this section.339 Operations that are ready to be submitted should be batched together in few system calls, but at the same time, operations should not be left pending for long period of times before being submitted. 340 Balancing submission can be handled by either designating one of the submitting \ats as the being responsible for the system call for the current batch of SQEs or by having some other party regularly submitting all ready SQEs, \eg, the poller \at mentioned later in this section. 303 341 304 342 Ideally, when multiple \ats attempt to submit operations to the same @io_uring@ instance, all requests should be batched together and one of the \ats is designated to do the system call on behalf of the others, called the \newterm{submitter}. 305 343 However, in practice, \io requests must be handed promptly so there is a need to guarantee everything missed by the current submitter is seen by the next one. 306 344 Indeed, as long as there is a ``next'' submitter, \ats submitting new \io requests can move on, knowing that some future system call includes their request. 307 Once the system call is done, the submitter must also free SQEs so that the allocator can reuse them.345 Once the system call is done, the submitter must also free SQEs so that the allocator can reused them. 308 346 309 347 Finally, the completion side is much simpler since the @io_uring@ system-call enforces a natural synchronization point. 310 348 Polling simply needs to regularly do the system call, go through the produced CQEs and communicate the result back to the originating \ats. 311 Since CQEs only own a signed 32 -bit result, in addition to the copy of the @user_data@ field, all that is needed to communicate the result is a simple future~\cite{wiki:future}.349 Since CQEs only own a signed 32 bit result, in addition to the copy of the @user_data@ field, all that is needed to communicate the result is a simple future~\cite{wiki:future}. 312 350 If the submission side does not designate submitters, polling can also submit all SQEs as it is polling events. 313 351 A simple approach to polling is to allocate a \at per @io_uring@ instance and simply let the poller \ats poll their respective instances when scheduled. … … 316 354 It does not impose restrictions on what \ats submitting \io operations can and cannot do between allocations and submissions. 317 355 It also can gracefully handle running out of resources, SQEs or the kernel returning @EBUSY@. 318 The down side to this approach is that many of the steps used for submitting need complex synchronization to work properly.356 The down side to this approach is that many of the steps used for submitting need complex synchronization to work properly. 319 357 The routing and allocation algorithm needs to keep track of which ring instances have available SQEs, block incoming requests if no instance is available, prevent barging if \ats are already queued up waiting for SQEs and handle SQEs being freed. 320 358 The submission side needs to safely append SQEs to the ring buffer, correctly handle chains, make sure no SQE is dropped or left pending forever, notify the allocation side when SQEs can be reused, and handle the kernel returning @EBUSY@. 321 Compared to the private-instance approach, all this synchronization has a significant costthis synchronization is entirely overhead.359 All this synchronization has a significant cost, and compared to the private-instance approach, this synchronization is entirely overhead. 322 360 323 361 \subsubsection{Instance borrowing} 324 362 Both of the prior approaches have undesirable aspects that stem from tight or loose coupling between @io_uring@ and \glspl{proc}. 325 363 The first approach suffers from tight coupling causing problems when a \gls{proc} does not benefit from the coupling. 326 The second approach suffers from loose coupling scausing operations to have synchronization overhead, which tighter coupling avoids.364 The second approach suffers from loose coupling causing operations to have synchronization overhead, which tighter coupling avoids. 327 365 When \glspl{proc} are continuously issuing \io operations, tight coupling is valuable since it avoids synchronization costs. 328 366 However, in unlikely failure cases or when \glspl{proc} are not using their instances, tight coupling is no longer advantageous. … … 334 372 When a \at attempts to issue an \io operation, it ask for an instance from the arbiter and issues requests to that instance. 335 373 This instance is now bound to the \gls{proc} the \at is running on. 336 This binding is kept until the arbiter decides to revoke it, taking back the instance and reverting the \gls{proc} to its initial \io state.374 This binding is kept until the arbiter decides to revoke it, taking back the instance and reverting the \gls{proc} to its initial state with respect to \io. 337 375 This tight coupling means that synchronization can be minimal since only one \gls{proc} can use the instance at a time, akin to the private instances approach. 338 376 However, it differs in that revocation by the arbiter means this approach does not suffer from the deadlock scenario described above. … … 345 383 \end{enumerate} 346 384 However, even when the arbiter is not directly needed, \glspl{proc} need to make sure that their instance ownership is not being revoked, which is accomplished by a lock-\emph{less} handshake.\footnote{ 347 Note the handshake is not lock -\emph{free} since it lacks the proper progress guarantee.}385 Note the handshake is not lock \emph{free} since it lacks the proper progress guarantee.} 348 386 A \gls{proc} raises a local flag before using its borrowed instance and checks if the instance is marked as revoked or if the arbiter has raised its flag. 349 387 If not, it proceeds, otherwise it delegates the operation to the arbiter. … … 351 389 352 390 Correspondingly, before revoking an instance, the arbiter marks the instance and then waits for the \gls{proc} using it to lower its local flag. 353 Only then does it reclaim the instance and potentially assign it to an other \gls{proc}.391 Only then does it reclaim the instance and potentially assign it to an other \gls{proc}. 354 392 355 393 The arbiter maintains four lists around which it makes its decisions: … … 372 410 373 411 While an arbiter has the potential to solve many of the problems mentioned above, it also introduces a significant amount of complexity. 374 Tracking which processors are borrowing which instances and which instances have SQEs available ends 412 Tracking which processors are borrowing which instances and which instances have SQEs available ends-up adding a significant synchronization prelude to any I/O operation. 375 413 Any submission must start with a handshake that pins the currently borrowed instance, if available. 376 414 An attempt to allocate is then made, but the arbiter can concurrently be attempting to allocate from the same instance from a different \gls{hthrd}. 377 415 Once the allocation is completed, the submission must check that the instance is still burrowed before attempting to flush. 378 416 These synchronization steps turn out to have a similar cost to the multiple shared-instances approach. 379 Furthermore, if the number of instances does not match the number of processors actively submitting I/O, the system can fall into a state where instances are constantly being revoked and end 417 Furthermore, if the number of instances does not match the number of processors actively submitting I/O, the system can fall into a state where instances are constantly being revoked and end-up cycling the processors, which leads to significant cache deterioration. 380 418 For these reasons, this approach, which sounds promising on paper, does not improve on the private instance approach in practice. 419 420 \subsubsection{Private Instances V2} 421 422 % Verbs of this design 423 424 % Allocation: obtaining an sqe from which to fill in the io request, enforces the io instance to use since it must be the one which provided the sqe. Must interact with the arbiter if the instance does not have enough sqe for the allocation. (Typical allocation will ask for only one sqe, but chained sqe must be allocated from the same context so chains of sqe must be allocated in bulks) 425 426 % Submission: simply adds the sqe(s) to some data structure to communicate that they are ready to go. This operation can't fail because there are as many spots in the submit buffer than there are sqes. Must interact with the arbiter only if the thread was moved between the allocation and the submission. 427 428 % Flushing: Taking all the sqes that were submitted and making them visible to the kernel, also counting them in order to figure out what to_submit should be. Must be thread-safe with submission. Has to interact with the Arbiter if there are external submissions. Can't simply use a protected queue because adding to the array is not safe if the ring is still available for submitters. Flushing must therefore: check if there are external pending requests if so, ask the arbiter to flush otherwise use the fast flush operation. 429 430 % Collect: Once the system call is done, it returns how many sqes were consumed by the system. These must be freed for allocation. Must interact with the arbiter to notify that things are now ready. 431 432 % Handle: process all the produced cqe. No need to interact with any of the submission operations or the arbiter. 433 434 435 % alloc(): 436 % proc.io->in_use = true, __ATOMIC_ACQUIRE 437 % if cltr.io.flag || !proc.io || proc.io->flag: 438 % return alloc_slow(cltr.io, proc.io) 439 440 % a = alloc_fast(proc.io) 441 % if a: 442 % proc.io->in_use = false, __ATOMIC_RELEASE 443 % return a 444 445 % return alloc_slow(cltr.io) 446 447 % alloc_fast() 448 % left = proc.io->submit_q.free.tail - proc.io->submit_q.free.head 449 % if num_entries - left < want: 450 % return None 451 452 % a = ready[head] 453 % head = head + 1, __ATOMIC_RELEASE 454 455 % alloc_slow() 456 % cltr.io.flag = true, __ATOMIC_ACQUIRE 457 % while(proc.io && proc.io->in_use) pause; 458 459 460 461 % submit(a): 462 % proc.io->in_use = true, __ATOMIC_ACQUIRE 463 % if cltr.io.flag || proc.io != alloc.io || proc.io->flag: 464 % return submit_slow(cltr.io) 465 466 % submit_fast(proc.io, a) 467 % proc.io->in_use = false, __ATOMIC_RELEASE 468 469 % polling() 470 % loop: 471 % yield 472 % flush() 473 % io_uring_enter 474 % collect 475 % handle() 381 476 382 477 \section{Interface} 383 478 The last important part of the \io subsystem is its interface. 384 Multiple approachescan be offered to programmers, each with advantages and disadvantages.479 There are multiple approaches that can be offered to programmers, each with advantages and disadvantages. 385 480 The new \io subsystem can replace the C runtime API or extend it, and in the later case, the interface can go from very similar to vastly different. 386 481 The following sections discuss some useful options using @read@ as an example. … … 394 489 The goal is to convince the compiler and linker to replace any calls to @read@ to direct them to the \CFA implementation instead of glibc's. 395 490 This rerouting has the advantage of working transparently and supporting existing binaries without needing recompilation. 396 It also offers a, presumably, well -known and familiar API that C programmers can simply continue to work with.491 It also offers a, presumably, well known and familiar API that C programmers can simply continue to work with. 397 492 However, this approach also entails a plethora of subtle technical challenges, which generally boils down to making a perfect replacement. 398 493 If the \CFA interface replaces only \emph{some} of the calls to glibc, then this can easily lead to esoteric concurrency bugs. 399 Since the gcc ecosystem does not offer a scheme for perfect replacement, this approach was rejected as being laudable but infeasible.494 Since the gcc ecosystems does not offer a scheme for perfect replacement, this approach was rejected as being laudable but infeasible. 400 495 401 496 \subsection{Synchronous Extension} … … 408 503 It comes with the caveat that any code attempting to use it must be recompiled, which is a problem considering the amount of existing legacy C binaries. 409 504 However, it has the advantage of implementation simplicity. 410 Finally, there is a certain irony to using a blocking synchronous interface for a feature often referred to as ``non-blocking'' \io.505 Finally, there is a certain irony to using a blocking synchronous interfaces for a feature often referred to as ``non-blocking'' \io. 411 506 412 507 \subsection{Asynchronous Extension} … … 436 531 This offers more flexibility to users wanting to fully utilize all of the @io_uring@ features. 437 532 However, it is not the most user-friendly option. 438 It obviously imposes a strong dependency between user code and @io_uring@ but at the same time restrict susers to usages that are compatible with how \CFA internally uses @io_uring@.533 It obviously imposes a strong dependency between user code and @io_uring@ but at the same time restricting users to usages that are compatible with how \CFA internally uses @io_uring@.
Note: See TracChangeset
for help on using the changeset viewer.