Files
lix/lix/libstore/filetransfer.cc
T
eldritch horrors f29dfb3d3c libutil: add buffer state management to loggers
currently all loggers can always accept messages and never suggest
flushing buffers. in the future this may change, and at that point
we're already fully set up for it. local loggers should never keep
asynchronous (i.e. network-backed) buffers, disk buffers are fine.
networked loggers will require buffers and periodic flushes later.

Change-Id: Ide2114f5bc17f4a1d289c92ed4f9381a1d59dacf
2025-10-13 11:26:49 +00:00

1227 lines
46 KiB
C++

#include "lix/libstore/filetransfer.hh"
#include "lix/libutil/async-io.hh"
#include "lix/libutil/async.hh"
#include "lix/libutil/c-calls.hh"
#include "lix/libutil/error.hh"
#include "lix/libutil/logging.hh"
#include "lix/libutil/namespaces.hh"
#include "lix/libstore/globals.hh"
#include "lix/libstore/store-api.hh"
#include "lix/libstore/s3.hh"
#include "lix/libutil/regex.hh"
#include "lix/libutil/result.hh"
#include "lix/libutil/signals.hh"
#include "lix/libutil/strings.hh"
#include "lix/libutil/thread-name.hh"
#include "lix/libutil/tracepoint.hh"
#include "lix/libutil/backoff.hh"
#include <cstddef>
#include <cstdio>
#include <exception>
#include <kj/async.h>
#include <kj/encoding.h>
#include <kj/time.h>
#if ENABLE_DTRACE
#include "trace-probes.gen.hh"
#endif
#if ENABLE_S3
#include <aws/core/client/ClientConfiguration.h>
#endif
#include <unistd.h>
#include <fcntl.h>
#include <curl/curl.h>
#include <algorithm>
#include <cmath>
#include <cstring>
#include <thread>
#include <regex>
using namespace std::string_literals;
namespace nix {
FileTransferSettings fileTransferSettings;
static GlobalConfig::Register rFileTransferSettings(&fileTransferSettings);
struct curlFileTransfer : public FileTransfer
{
std::unique_ptr<CURLM, decltype([](auto * m) { curl_multi_cleanup(m); })> curlm;
const unsigned int baseRetryTimeMs;
struct TransferItem
{
struct DownloadState
{
bool done = false;
std::exception_ptr exc;
std::string data;
std::optional<kj::Own<kj::CrossThreadPromiseFulfiller<void>>> downloadEvent;
auto wait()
{
auto pfp = kj::newPromiseAndCrossThreadFulfiller<void>();
downloadEvent = std::move(pfp.fulfiller);
return std::move(pfp.promise);
}
void signal()
{
if (downloadEvent) {
(*downloadEvent)->fulfill();
downloadEvent.reset();
}
}
};
std::string uri;
FileTransferResult result;
Activity act;
std::unique_ptr<FILE, decltype([](FILE * f) { fclose(f); })> uploadData;
Sync<DownloadState> downloadState;
bool headersDone = false, metadataReturned = false;
kj::Own<kj::CrossThreadPromiseFulfiller<Result<FileTransferResult>>> metadataPromise;
std::string statusMsg;
uint64_t bodySize = 0;
std::unique_ptr<curl_slist, decltype([](auto * s) { curl_slist_free_all(s); })> requestHeaders;
std::unique_ptr<CURL, decltype([](auto * c) { curl_easy_cleanup(c); })> req;
// buffer to accompany the `req` above
char errbuf[CURL_ERROR_SIZE];
inline static const std::set<long> successfulStatuses {200, 201, 204, 206, 304, 0 /* other protocol */};
std::optional<long> httpStatusCode;
/* Get the scheme for the current curl handle, or none if curl returns NULL.
* Ensures the scheme is always casefolded to lowercase */
std::optional<std::string> getCurlScheme() {
char *scheme_raw = nullptr;
if (curl_easy_getinfo(req.get(), CURLINFO_SCHEME, &scheme_raw) != CURLE_OK) {
throw nix::Error("could not get scheme used from curl handle");
}
if (scheme_raw) {
return toLower(std::string(scheme_raw));
} else {
return {};
}
}
/* Get the HTTP status code, or 0 for other protocols. */
long getHTTPStatus()
{
if (httpStatusCode) {
return *httpStatusCode;
}
long statusCode = 0;
std::optional<std::string> scheme = getCurlScheme();
if (scheme == "http" || scheme == "https") {
if (curl_easy_getinfo(req.get(), CURLINFO_RESPONSE_CODE, &statusCode) != CURLE_OK) {
throw nix::Error("could not get response code from curl handle");
}
}
httpStatusCode = statusCode;
return statusCode;
}
std::string verb() const
{
return uploadData ? "upload" : "download";
}
TransferItem(
const std::string & uri,
FileTransferOptions && options,
const Activity * parentAct,
std::optional<std::string_view> uploadData,
bool noBody,
curl_off_t writtenToSink,
kj::Own<kj::CrossThreadPromiseFulfiller<Result<FileTransferResult>>> metadataPromise,
const std::chrono::milliseconds & connectTimeout
)
: uri(uri)
, act(logger->startActivity(
lvlTalkative,
actFileTransfer,
fmt(uploadData ? "uploading '%s'" : "downloading '%s'", uri),
{uri},
parentAct
))
, metadataPromise(std::move(metadataPromise))
, req(curl_easy_init())
{
if (req == nullptr) {
throw FileTransferError(Misc, {}, "could not allocate curl handle");
}
for (auto it = options.headers.begin(); it != options.headers.end(); ++it){
if (auto next = curl_slist_append(
requestHeaders.get(), requireCString(fmt("%s: %s", it->first, it->second))
);
next != nullptr)
{
(void) requestHeaders.release(); // next now owns this pointer
requestHeaders.reset(next);
} else {
throw FileTransferError(Misc, {}, "could not allocate curl request headers");
}
}
if (verbosity >= lvlVomit) {
curl_easy_setopt(req.get(), CURLOPT_VERBOSE, 1);
curl_easy_setopt(req.get(), CURLOPT_DEBUGFUNCTION, TransferItem::debugCallback);
}
curl_easy_setopt(req.get(), CURLOPT_URL, uri.c_str());
curl_easy_setopt(req.get(), CURLOPT_FOLLOWLOCATION, 1L);
curl_easy_setopt(req.get(), CURLOPT_ACCEPT_ENCODING, ""); // all of them!
curl_easy_setopt(req.get(), CURLOPT_MAXREDIRS, 10);
curl_easy_setopt(req.get(), CURLOPT_NOSIGNAL, 1);
curl_easy_setopt(req.get(), CURLOPT_USERAGENT,
("curl/" LIBCURL_VERSION " Lix/" + nixVersion +
(fileTransferSettings.userAgentSuffix != "" ? " " + fileTransferSettings.userAgentSuffix.get() : "")).c_str());
curl_easy_setopt(req.get(), CURLOPT_PIPEWAIT, 1);
if (fileTransferSettings.enableHttp2)
curl_easy_setopt(req.get(), CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_2TLS);
else
curl_easy_setopt(req.get(), CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_1_1);
curl_easy_setopt(req.get(), CURLOPT_WRITEFUNCTION, TransferItem::writeCallbackWrapper);
curl_easy_setopt(req.get(), CURLOPT_WRITEDATA, this);
curl_easy_setopt(req.get(), CURLOPT_HEADERFUNCTION, TransferItem::headerCallbackWrapper);
curl_easy_setopt(req.get(), CURLOPT_HEADERDATA, this);
curl_easy_setopt(req.get(), CURLOPT_XFERINFOFUNCTION, progressCallbackWrapper);
curl_easy_setopt(req.get(), CURLOPT_PROGRESSDATA, this);
curl_easy_setopt(req.get(), CURLOPT_NOPROGRESS, 0);
curl_easy_setopt(req.get(), CURLOPT_ERRORBUFFER, errbuf);
errbuf[0] = 0;
curl_easy_setopt(req.get(), CURLOPT_PROTOCOLS_STR, "http,https,ftp,ftps");
curl_easy_setopt(req.get(), CURLOPT_HTTPHEADER, requestHeaders.get());
if (settings.downloadSpeed.get() > 0)
curl_easy_setopt(req.get(), CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t) (settings.downloadSpeed.get() * 1024));
if (noBody)
curl_easy_setopt(req.get(), CURLOPT_NOBODY, 1);
if (uploadData) {
this->uploadData.reset(fmemopen(const_cast<char *>(uploadData->data()), uploadData->size(), "r"));
curl_easy_setopt(req.get(), CURLOPT_UPLOAD, 1L);
curl_easy_setopt(req.get(), CURLOPT_READDATA, this->uploadData.get());
curl_easy_setopt(req.get(), CURLOPT_INFILESIZE_LARGE, (curl_off_t) uploadData->length());
}
if (settings.caFile != "")
curl_easy_setopt(req.get(), CURLOPT_CAINFO, settings.caFile.get().c_str());
curl_easy_setopt(req.get(), CURLOPT_CONNECTTIMEOUT_MS, connectTimeout.count());
curl_easy_setopt(req.get(), CURLOPT_LOW_SPEED_LIMIT, 1L);
curl_easy_setopt(req.get(), CURLOPT_LOW_SPEED_TIME, fileTransferSettings.stalledDownloadTimeout.get());
/* If no file exist in the specified path, curl continues to work
anyway as if netrc support was disabled. */
curl_easy_setopt(req.get(), CURLOPT_NETRC_FILE, settings.netrcFile.get().c_str());
curl_easy_setopt(req.get(), CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
if (writtenToSink)
curl_easy_setopt(req.get(), CURLOPT_RESUME_FROM_LARGE, writtenToSink);
if (options.extraSetup) {
options.extraSetup(req.get());
}
}
bool acceptsRanges()
{
curl_header * h;
if (curl_easy_header(req.get(), "accept-ranges", 0, CURLH_HEADER, -1, &h)) {
// treat any error as the remote not accepting range requests. the only
// interesting local error is out-of-memory, which we can't even handle
return false;
}
return toLower(trim(h->value)) == "bytes";
}
void failEx(std::exception_ptr ex)
{
auto state = downloadState.lock();
assert(!state->done && !state->exc);
if (!metadataReturned) {
metadataPromise->fulfill(ex);
}
state->exc = ex;
state->signal();
}
template<class T>
void fail(T && e)
{
failEx(std::make_exception_ptr(std::forward<T>(e)));
}
void maybeFinishSetup()
{
if (headersDone) {
return;
}
auto status = getHTTPStatus();
char * effectiveUriCStr = nullptr;
curl_easy_getinfo(req.get(), CURLINFO_EFFECTIVE_URL, &effectiveUriCStr);
if (effectiveUriCStr) {
result.effectiveUri = effectiveUriCStr;
}
result.cached = status == 304;
if (successfulStatuses.contains(status)) {
metadataPromise->fulfill(result);
metadataReturned = true;
}
headersDone = true;
}
std::exception_ptr callbackException;
size_t writeCallback(void * contents, size_t size, size_t nmemb)
{
const size_t realSize = size * nmemb;
try {
maybeFinishSetup();
auto state = downloadState.lock();
// when the buffer is full (as determined by a historical magic value) we
// pause the transfer and wait for the receiver to unpause it when ready.
if (successfulStatuses.count(getHTTPStatus()) && state->data.size() > 1024 * 1024) {
return CURL_WRITEFUNC_PAUSE;
}
state->data.append(static_cast<const char *>(contents), realSize);
state->signal();
bodySize += realSize;
return realSize;
} catch (...) {
callbackException = std::current_exception();
return CURL_WRITEFUNC_ERROR;
}
}
static size_t writeCallbackWrapper(void * contents, size_t size, size_t nmemb, void * userp)
{
return static_cast<TransferItem *>(userp)->writeCallback(contents, size, nmemb);
}
size_t headerCallback(void * contents, size_t size, size_t nmemb)
try {
size_t realSize = size * nmemb;
std::string line(static_cast<char *>(contents), realSize);
printMsg(lvlVomit, "got header for '%s': %s", uri, trim(line));
static std::regex statusLine = regex::parse("HTTP/[^ ]+ +[0-9]+(.*)", std::regex::extended | std::regex::icase);
if (std::smatch match; std::regex_match(line, match, statusLine)) {
statusMsg = trim(match.str(1));
} else {
auto i = line.find(':');
if (i != std::string::npos) {
std::string name = toLower(trim(line.substr(0, i)));
if (name == "etag") {
// NOTE we don't check that the etag hasn't gone *missing*. technically
// this is not an error as long as we get the same data from the remote.
auto etag = trim(line.substr(i + 1));
result.etag = std::move(etag);
}
else if (name == "link" || name == "x-amz-meta-link") {
auto value = trim(line.substr(i + 1));
static std::regex linkRegex = regex::parse("<([^>]*)>; rel=\"immutable\"", std::regex::extended | std::regex::icase);
if (std::smatch match; std::regex_match(value, match, linkRegex)) {
result.immutableUrl = match.str(1);
} else
debug("got invalid link header '%s'", value);
}
}
}
return realSize;
} catch (...) {
callbackException = std::current_exception();
return CURL_WRITEFUNC_ERROR;
}
static size_t headerCallbackWrapper(void * contents, size_t size, size_t nmemb, void * userp)
{
return static_cast<TransferItem *>(userp)->headerCallback(contents, size, nmemb);
}
int progressCallback(curl_off_t dltotal, curl_off_t dlnow)
{
try {
if (act.progress(dlnow, dltotal) == Logger::BufferState::NeedsFlush) {
act.getLogger().waitForSpace();
}
} catch (nix::Interrupted &) {
}
return isInterrupted();
}
static int progressCallbackWrapper(void * userp, curl_off_t dltotal, curl_off_t dlnow, curl_off_t ultotal, curl_off_t ulnow)
{
return static_cast<TransferItem *>(userp)->progressCallback(dltotal, dlnow);
}
static int debugCallback(CURL * handle, curl_infotype type, char * data, size_t size, void * userptr)
{
if (type == CURLINFO_TEXT)
vomit("curl: %s", chomp(std::string(data, size)));
return 0;
}
void finish(CURLcode code)
{
auto httpStatus = getHTTPStatus();
maybeFinishSetup();
debug("finished %s of '%s'; curl status = %d, HTTP status = %d, body = %d bytes",
verb(), uri, code, httpStatus, bodySize);
if (callbackException)
failEx(callbackException);
else if (code == CURLE_OK && successfulStatuses.count(httpStatus))
{
if (act.progress(bodySize, bodySize) == Logger::BufferState::NeedsFlush) {
act.getLogger().waitForSpace();
}
auto state = downloadState.lock();
state->done = true;
state->signal();
}
else {
// We treat most errors as transient, but won't retry when hopeless
Error err = Transient;
if (httpStatus == 404 || httpStatus == 410 || code == CURLE_FILE_COULDNT_READ_FILE) {
// The file is definitely not there
err = NotFound;
} else if (httpStatus == 401 || httpStatus == 403 || httpStatus == 407) {
// Don't retry on authentication/authorization failures
err = Forbidden;
} else if (httpStatus >= 400 && httpStatus < 500 && httpStatus != 408 && httpStatus != 429) {
// Most 4xx errors are client errors and are probably not worth retrying:
// * 408 means the server timed out waiting for us, so we try again
// * 429 means too many requests, so we retry (with a delay)
err = Misc;
} else if (httpStatus == 501 || httpStatus == 505 || httpStatus == 511) {
// Let's treat most 5xx (server) errors as transient, except for a handful:
// * 501 not implemented
// * 505 http version not supported
// * 511 we're behind a captive portal
err = Misc;
} else {
// Don't bother retrying on certain cURL errors either
// Allow selecting a subset of enum values
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wswitch-enum"
switch (code) {
case CURLE_FAILED_INIT:
case CURLE_URL_MALFORMAT:
case CURLE_NOT_BUILT_IN:
case CURLE_REMOTE_ACCESS_DENIED:
case CURLE_FILE_COULDNT_READ_FILE:
case CURLE_FUNCTION_NOT_FOUND:
case CURLE_ABORTED_BY_CALLBACK:
case CURLE_BAD_FUNCTION_ARGUMENT:
case CURLE_INTERFACE_FAILED:
case CURLE_UNKNOWN_OPTION:
case CURLE_SSL_CACERT_BADFILE:
case CURLE_TOO_MANY_REDIRECTS:
case CURLE_WRITE_ERROR:
case CURLE_UNSUPPORTED_PROTOCOL:
err = Misc;
break;
default: // Shut up warnings
break;
}
#pragma GCC diagnostic pop
}
std::optional<std::string> response;
if (!successfulStatuses.count(httpStatus))
response = std::move(downloadState.lock()->data);
auto textualError = [](const char * errbuf, CURLcode code) -> const char * {
if (errbuf && errbuf[0]) {
return errbuf;
} else {
return curl_easy_strerror(code);
}
};
auto exc = code == CURLE_ABORTED_BY_CALLBACK && isInterrupted()
? FileTransferError(
Interrupted,
std::move(response),
"%s of '%s' was interrupted",
verb(),
uri
)
: httpStatus != 0
? FileTransferError(
err,
std::move(response),
"unable to %s '%s': HTTP error %d (%s)%s",
verb(),
uri,
httpStatus,
statusMsg,
code == CURLE_OK
? ""
: fmt(" (curl error code=%d: %s)", code, textualError(errbuf, code))
)
: FileTransferError(
err,
std::move(response),
"unable to %s '%s': %s (curl error code=%d)",
verb(),
uri,
textualError(errbuf, code),
code
);
fail(std::move(exc));
}
}
};
void unpause(const std::shared_ptr<TransferItem> & transfer)
{
auto lock = state_.lock();
lock->unpause.push_back(transfer);
wakeup();
}
void cancel(const std::shared_ptr<TransferItem> & transfer)
{
std::promise<void> promise;
auto wait = promise.get_future();
{
auto lock = state_.lock();
if (lock->quit) {
return;
}
lock->cancel[transfer] = std::move(promise);
}
wakeup();
wait.get();
}
struct State
{
bool quit = false;
std::vector<std::shared_ptr<TransferItem>> incoming;
std::vector<std::shared_ptr<TransferItem>> unpause;
std::map<std::shared_ptr<TransferItem>, std::promise<void>> cancel;
};
Sync<State> state_;
std::thread workerThread;
curlFileTransfer(unsigned int baseRetryTimeMs)
: curlm(curl_multi_init())
, baseRetryTimeMs(baseRetryTimeMs)
{
if (curlm == nullptr) {
throw FileTransferError(Misc, {}, "could not allocate curl handle");
}
static std::once_flag globalInit;
std::call_once(globalInit, curl_global_init, CURL_GLOBAL_ALL);
curl_multi_setopt(curlm.get(), CURLMOPT_PIPELINING, CURLPIPE_MULTIPLEX);
curl_multi_setopt(curlm.get(), CURLMOPT_MAX_TOTAL_CONNECTIONS,
fileTransferSettings.httpConnections.get());
workerThread = std::thread([&]() {
setCurrentThreadName("curlFileTransfer worker");
workerThreadEntry();
});
}
~curlFileTransfer()
{
try {
stopWorkerThread();
} catch (nix::Error & e) {
// This can only fail if a socket to our own process cannot be
// written to, so it is always a bug in the program if it fails.
//
// Joining the thread would probably only cause a deadlock if this
// happened, so just die on purpose.
printError("failed to join curl file transfer worker thread: %1%", e.what());
std::terminate();
}
workerThread.join();
}
void wakeup()
{
if (auto mc = curl_multi_wakeup(curlm.get()))
throw nix::Error("unexpected error from curl_multi_wakeup(): %s", curl_multi_strerror(mc));
}
void stopWorkerThread()
{
/* Signal the worker thread to exit. */
{
auto state(state_.lock());
state->quit = true;
}
wakeup();
}
void workerThreadMain()
{
/* Cause this thread to be notified on SIGINT. */
auto callback = createInterruptCallback([&]() {
stopWorkerThread();
});
unshareFilesystem();
std::map<CURL *, std::shared_ptr<TransferItem>> items;
// clear all current transfers in case of an early exit, as can happen
// via Interrupted if the interruption occured right before a log call
KJ_DEFER({
for (auto & [_, item] : items) {
item->finish(CURLE_ABORTED_BY_CALLBACK);
}
});
bool quit = false;
// NOTE: we will need to use CURLMOPT_TIMERFUNCTION to integrate this
// loop with kj. until then curl will handle its timeouts internally.
int64_t timeoutMs = INT64_MAX;
while (true) {
{
auto cancel = [&] { return std::move(state_.lock()->cancel); }();
for (auto & [item, promise] : cancel) {
curl_multi_remove_handle(curlm.get(), item->req.get());
items.erase(item->req.get());
promise.set_value();
}
}
/* Let curl do its thing. */
int running;
CURLMcode mc = curl_multi_perform(curlm.get(), &running);
if (mc != CURLM_OK)
throw nix::Error("unexpected error from curl_multi_perform(): %s", curl_multi_strerror(mc));
/* Set the promises of any finished requests. */
CURLMsg * msg;
int left;
while ((msg = curl_multi_info_read(curlm.get(), &left))) {
if (msg->msg == CURLMSG_DONE) {
auto i = items.find(msg->easy_handle);
assert(i != items.end());
i->second->finish(msg->data.result);
curl_multi_remove_handle(curlm.get(), i->second->req.get());
items.erase(i);
}
}
// only exit when all transfers are done (which will happen through the
// progress callback issuing an abort in the case of user interruption)
if (items.empty() && quit) {
break;
}
/* Wait for activity, including wakeup events. */
mc = curl_multi_poll(curlm.get(), nullptr, 0, std::min<int64_t>(timeoutMs, INT_MAX), nullptr);
if (mc != CURLM_OK)
throw nix::Error("unexpected error from curl_multi_poll(): %s", curl_multi_strerror(mc));
/* Add new curl requests from the incoming requests queue,
except for requests that are embargoed (waiting for a
retry timeout to expire). */
std::vector<std::shared_ptr<TransferItem>> incoming;
timeoutMs = INT64_MAX;
{
auto unpause = [&] { return std::move(state_.lock()->unpause); }();
for (auto & item : unpause) {
curl_easy_pause(item->req.get(), CURLPAUSE_CONT);
}
}
{
auto state(state_.lock());
incoming = std::move(state->incoming);
quit = state->quit;
}
for (auto & item : incoming) {
debug("starting %s of %s", item->verb(), item->uri);
curl_multi_add_handle(curlm.get(), item->req.get());
items[item->req.get()] = item;
}
}
debug("download thread shutting down");
}
void workerThreadEntry()
{
try {
workerThreadMain();
} catch (nix::Interrupted & e) {
} catch (std::exception & e) { // NOLINT(lix-foreign-exceptions)
printError("unexpected error in download thread: %s", e.what());
} catch (...) {
printError("unexpected error in download thread");
}
{
auto state(state_.lock());
for (auto & item : state->incoming) {
item->finish(CURLE_ABORTED_BY_CALLBACK);
}
state->incoming.clear();
state->quit = true;
}
}
void enqueueItem(std::shared_ptr<TransferItem> item)
{
if (item->uploadData
&& !item->uri.starts_with("http://")
&& !item->uri.starts_with("https://"))
throw nix::Error("uploading to '%s' is not supported", item->uri);
{
auto state(state_.lock());
if (state->quit)
throw nix::Error("cannot enqueue download request because the download thread is shutting down");
state->incoming.push_back(item);
}
wakeup();
}
#if ENABLE_S3
static std::tuple<std::string, std::string, StoreConfig::Params> parseS3Uri(std::string uri)
{
auto [path, params] = splitUriAndParams(uri);
auto slash = path.find('/', 5); // 5 is the length of "s3://" prefix
if (slash == std::string::npos)
throw nix::Error("bad S3 URI '%s'", path);
std::string bucketName(path, 5, slash - 5);
std::string key(path, slash + 1);
return {bucketName, key, params};
}
#endif
kj::Promise<Result<void>> upload(
const std::string & uri,
std::string data,
FileTransferOptions options,
const Activity * context
) override
try {
TRY_AWAIT(enqueueFileTransfer(uri, std::move(options), std::move(data), false, context));
co_return result::success();
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<std::optional<std::pair<FileTransferResult, box_ptr<AsyncInputStream>>>>>
tryEagerTransfers(
const std::string & uri,
const FileTransferOptions & options,
const std::optional<std::string> & data,
bool noBody
)
try {
// curl transfers using file:// urls cannot be paused, and are a bit unruly
// in other ways too. since their metadata is trivial and we already have a
// backend for simple file system reads we can use that instead. we'll pass
// uploads to files to curl even so, those will fail in enqueueItem anyway.
// on all other decoding failures we also let curl fail for us a bit later.
//
// note that we use kj to decode the url, not curl. curl uses only the path
// component of the url to determine the file name, but it does note expose
// the decoding method it uses for this. for file:// transfers curl forbids
// only \0 characters in the urldecoded path, not all control characters as
// it does in the public curl_url_get(CURLUPART_PATH, CURLU_URLDECODE) api.
//
// also note: everything weird you see here is for compatibility with curl.
// we can't even fix it because nix-channel relies on this. even reading of
// directories being allowed and returning something (though hopefully it's
// enough to return anything instead of a directory listing like curl does)
if (uri.starts_with("file://") && !data.has_value()) {
if (!uri.starts_with("file:///")) {
throw FileTransferError(NotFound, std::nullopt, "file not found");
}
auto url = curl_url();
if (!url) {
throw std::bad_alloc();
}
KJ_DEFER(curl_url_cleanup(url));
curl_url_set(url, CURLUPART_URL, requireCString(uri), 0);
char * path = nullptr;
curl_url_get(url, CURLUPART_PATH, &path, 0);
auto decoded = kj::decodeUriComponent(kj::arrayPtr(path, path + strlen(path)));
if (!decoded.hadErrors && decoded.findFirst(0) == nullptr) {
Path fsPath(decoded.cStr(), decoded.size());
FileTransferResult metadata{.effectiveUri = std::string("file://") + path};
struct stat st;
AutoCloseFD fd(sys::open(fsPath, O_RDONLY));
if (!fd || fstat(fd.get(), &st) != 0) {
throw FileTransferError(
NotFound, std::nullopt, "%s: file not found (%s)", fsPath, strerror(errno)
);
}
if (S_ISDIR(st.st_mode)) {
co_return std::pair{
std::move(metadata), make_box_ptr<AsyncStringInputStream>("")
};
}
struct OwningFdStream : AsyncInputStream
{
AutoCloseFD fd;
OwningFdStream(AutoCloseFD fd) : fd(std::move(fd)) {}
kj::Promise<Result<std::optional<size_t>>>
read(void * buffer, size_t size) override
{
// NOTE the synchronous implementation used to have a buffer for
// file data, but we cannot be bothered to treat this edge case.
if (const auto got = ::read(fd.get(), buffer, size); got >= 0) {
if (got == 0) {
return {result::success(std::nullopt)};
} else {
return {result::success(got)};
}
} else {
return {result::failure(std::make_exception_ptr(SysError("reading file")
))};
}
}
};
co_return std::pair{
std::move(metadata), make_box_ptr<OwningFdStream>(std::move(fd))
};
}
}
/* Ugly hack to support s3:// URIs. */
if (uri.starts_with("s3://")) {
// FIXME: do this on a worker thread
#if ENABLE_S3
auto [bucketName, key, params] = parseS3Uri(uri);
std::string profile = getOr(params, "profile", "");
std::string region = getOr(params, "region", Aws::Region::US_EAST_1);
std::string scheme = getOr(params, "scheme", "");
std::string endpoint = getOr(params, "endpoint", "");
S3Helper s3Helper(profile, region, scheme, endpoint);
// FIXME: implement ETag
auto s3Res = TRY_AWAIT(s3Helper.getObject(bucketName, key));
FileTransferResult res;
if (!s3Res.data)
throw FileTransferError(NotFound, "S3 object '%s' does not exist", uri);
struct OwningStringStream : private std::string, AsyncStringInputStream
{
OwningStringStream(std::string data)
: std::string(std::move(data))
, AsyncStringInputStream(*this)
{
}
};
co_return std::pair{res, make_box_ptr<OwningStringStream>(std::move(*s3Res.data))};
#else
throw nix::Error(
"cannot download '%s' because Lix is not built with S3 support", uri
);
#endif
}
co_return std::nullopt;
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<std::pair<FileTransferResult, box_ptr<AsyncInputStream>>>>
enqueueFileTransfer(
const std::string & uri,
FileTransferOptions && options,
std::optional<std::string> data,
bool noBody,
const Activity * context
)
try {
if (auto eager = TRY_AWAIT(tryEagerTransfers(uri, options, data, noBody))) {
co_return std::move(*eager);
}
auto source = make_box_ptr<TransferStream>(
*this, uri, std::move(options), std::move(data), noBody, context
);
TRY_AWAIT(source->init());
TRY_AWAIT(source->awaitData());
co_return {source->metadata, std::move(source)};
} catch (...) {
co_return result::current_exception();
}
struct TransferStream : AsyncInputStream
{
curlFileTransfer & parent;
std::string uri;
FileTransferOptions options;
std::optional<std::string> data;
bool noBody;
const Activity * parentAct;
std::shared_ptr<TransferItem> transfer;
FileTransferResult metadata;
std::string chunk;
std::string_view buffered;
const size_t tries = fileTransferSettings.tries;
curl_off_t totalReceived = 0;
Generator<BackoffTiming> backoff;
TransferStream(
curlFileTransfer & parent,
const std::string & uri,
FileTransferOptions && options,
std::optional<std::string> data,
bool noBody,
const Activity * context
)
: parent(parent)
, uri(uri)
, options(options)
, data(std::move(data))
, noBody(noBody)
, parentAct(context)
, backoff(backoffTimeouts(
fileTransferSettings.tries,
std::chrono::seconds(fileTransferSettings.maxConnectTimeout.get()),
std::chrono::seconds(fileTransferSettings.initialConnectTimeout.get()),
std::chrono::milliseconds(parent.baseRetryTimeMs)
))
{
}
~TransferStream()
{
// wake up the download thread if it's still going and have it abort
try {
if (transfer) {
parent.cancel(transfer);
}
} catch (...) {
ignoreExceptionInDestructor();
}
}
kj::Promise<Result<void>> init()
try {
metadata = TRY_AWAIT(withRetries(
[&]() {
return startTransfer(
uri, std::chrono::seconds(fileTransferSettings.initialConnectTimeout.get())
);
},
[&](const std::chrono::milliseconds & timeout) {
return startTransfer(uri, timeout);
}
));
co_return result::success();
} catch (...) {
co_return result::current_exception();
}
auto withRetries(auto && initial, auto && retry) -> decltype(initial())
try {
std::optional<std::string> retryContext;
BackoffTiming timings;
while (true) {
try {
if (retryContext) {
TRY_AWAIT(prepareRetry(*retryContext, timings.waitTime, timings.attempt));
co_return TRY_AWAIT(retry(timings.downloadTimeout));
} else {
co_return TRY_AWAIT(initial());
}
} catch (FileTransferError & e) {
auto next = backoff.next();
// If this is a transient error, then maybe retry after a while. after any
// bytes have been received we require range support to proceed, otherwise
// we'd need to start from scratch and discard everything we already have.
if (e.error != Transient || data.has_value() || !next.has_value()
|| (totalReceived > 0 && !transfer->acceptsRanges()))
{
throw;
}
retryContext = e.what();
timings = *next;
}
}
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<FileTransferResult>> startTransfer(
const std::string & uri,
const std::chrono::milliseconds & timeout,
curl_off_t offset = 0
)
try {
auto uploadData = data ? std::optional(std::string_view(*data)) : std::nullopt;
auto pfp = kj::newPromiseAndCrossThreadFulfiller<Result<FileTransferResult>>();
transfer = std::make_shared<TransferItem>(
uri,
std::move(options),
parentAct,
uploadData,
noBody,
offset,
std::move(pfp.fulfiller),
timeout
);
parent.enqueueItem(transfer);
co_return TRY_AWAIT(pfp.promise);
} catch (...) {
co_return result::current_exception();
}
void throwChangedTarget(std::string_view what, std::string_view from, std::string_view to)
{
if (!from.empty() && from != to) {
throw FileTransferError(
Misc, {}, "uri %s changed %s from %s to %s during transfer", uri, what, from, to
);
}
}
kj::Promise<Result<void>> prepareRetry(
const std::string & context,
const std::chrono::milliseconds & waitTime,
unsigned int attempt
)
try {
if (totalReceived) {
printTaggedWarning(
"%s; retrying from offset %d in %d ms (attempt %d/%d)",
Uncolored(context),
totalReceived,
waitTime,
Uncolored(attempt),
Uncolored(tries)
);
} else {
printTaggedWarning(
"%s; retrying in %d ms (attempt %d/%d)",
Uncolored(context),
waitTime,
Uncolored(attempt),
Uncolored(tries)
);
}
co_await AIO().provider.getTimer().afterDelay(waitTime.count() * kj::MILLISECONDS);
co_return result::success();
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<void>> restartTransfer(const std::chrono::milliseconds & timeout)
try {
// use the effective URI of the previous transfer for retries. this avoids
// some silent corruption if a redirect changes between starting and retry
const auto & uri = metadata.effectiveUri.empty() ? this->uri : metadata.effectiveUri;
auto newMeta = TRY_AWAIT(startTransfer(uri, timeout, totalReceived));
throwChangedTarget("final destination", metadata.effectiveUri, newMeta.effectiveUri);
throwChangedTarget("ETag", metadata.etag, newMeta.etag);
throwChangedTarget(
"immutable url",
metadata.immutableUrl.value_or(""),
newMeta.immutableUrl.value_or("")
);
co_return result::success();
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<bool>> waitForData()
try {
/* Grab data if available, otherwise wait for the download
thread to wake us up. */
std::optional<kj::Promise<void>> signal;
while (buffered.empty()) {
if (signal) {
co_await *signal;
signal.reset();
}
auto state(transfer->downloadState.lock());
if (!state->data.empty()) {
chunk = std::move(state->data);
buffered = chunk;
totalReceived += chunk.size();
parent.unpause(transfer);
} else if (state->exc) {
std::rethrow_exception(state->exc);
} else if (state->done) {
co_return false;
} else {
parent.unpause(transfer);
signal = state->wait();
}
}
co_return true;
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<bool>> restartAndWaitForData(const std::chrono::milliseconds & timeout)
try {
TRY_AWAIT(restartTransfer(timeout));
co_return TRY_AWAIT(waitForData());
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<bool>> awaitData()
try {
co_return TRY_AWAIT(withRetries(
[&] { return waitForData(); },
[&](const std::chrono::milliseconds & timeout) {
return restartAndWaitForData(timeout);
}
));
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<std::optional<size_t>>> read(void * buffer, size_t len) override
try {
TRACE(LIX_STORE_FILETRANSFER_READ(uri.c_str(), len));
size_t total = 0;
auto data = static_cast<char *>(buffer);
while (total < len && TRY_AWAIT(awaitData())) {
const auto available = std::min(len - total, buffered.size());
memcpy(data + total, buffered.data(), available);
buffered.remove_prefix(available);
total += available;
}
if (total == 0) {
co_return std::nullopt;
} else {
co_return total;
}
} catch (...) {
co_return result::current_exception();
}
};
kj::Promise<Result<bool>>
exists(const std::string & uri, FileTransferOptions options, const Activity * context) override
try {
try {
TRY_AWAIT(enqueueFileTransfer(uri, std::move(options), std::nullopt, true, context));
co_return true;
} catch (FileTransferError & e) {
/* S3 buckets return 403 if a file doesn't exist and the
bucket is unlistable, so treat 403 as 404. */
if (e.error == FileTransfer::NotFound || e.error == FileTransfer::Forbidden)
co_return false;
throw;
}
} catch (...) {
co_return result::current_exception();
}
kj::Promise<Result<std::pair<FileTransferResult, box_ptr<AsyncInputStream>>>> download(
const std::string & uri, FileTransferOptions options, const Activity * context
) override
{
return enqueueFileTransfer(uri, std::move(options), std::nullopt, false, context);
}
};
ref<curlFileTransfer> makeCurlFileTransfer(std::optional<unsigned int> baseRetryTimeMs)
{
return make_ref<curlFileTransfer>(baseRetryTimeMs.value_or(250));
}
ref<FileTransfer> getFileTransfer()
{
static ref<curlFileTransfer> fileTransfer = makeCurlFileTransfer({});
if (fileTransfer->state_.lock()->quit)
fileTransfer = makeCurlFileTransfer({});
return fileTransfer;
}
ref<FileTransfer> makeFileTransfer(std::optional<unsigned int> baseRetryTimeMs)
{
return makeCurlFileTransfer(baseRetryTimeMs);
}
template<typename... Args>
FileTransferError::FileTransferError(FileTransfer::Error error, std::optional<std::string> response, const Args & ... args)
: Error(args...), error(error), response(response)
{
const auto hf = HintFmt(args...);
// FIXME: Due to https://github.com/NixOS/nix/issues/3841 we don't know how
// to print different messages for different verbosity levels. For now
// we add some heuristics for detecting when we want to show the response.
if (response && (response->size() < 1024 || response->find("<html>") != std::string::npos))
err.msg = HintFmt("%1%\n\nresponse body:\n\n%2%", Uncolored(hf.str()), chomp(*response));
else
err.msg = hf;
}
}