Cache servers differ from regular servers in that they do not produce
warnings and are not removed from the server pool for "soft errors"
(i.e. the server was reachable, but the download failed) and they are
not used for databases. If a host is used for both a cache server and a
regular server, it may still be removed from the server pool for soft
errors that occur when used as cache server and removal from the server
pool for soft errors will not affect future attempted use as a cache
server.
Signed-off-by: Andrew Gregory <andrew.gregory.8@gmail.com>
We have not set handle in the function at this stage, so we can not
assign an error to it. Pass the handle to the function to avoid
waiting until the payload is retrieved.
Signed-off-by: Allan McRae <allan@archlinux.org>
The last user of ABORT_SIGINT was removed in commit 84723cab5d
("Cleanup the old sequential download code"), and this isn't exported as
part of the public API.
Signed-off-by: Chris Down <chris@chrisdown.name>
Signed-off-by: Allan McRae <allan@archlinux.org>
Parsing of Content-Disposition relies on well formed headers.
A malformed header such as:
Content-Disposition="";
will result in a strnduppayload->content_disp_name, -1, ptr),
which will copy memory until it hits a \0.
Prevent this by only copying the value if it exists.
Fixes FS#73704.
Signed-off-by: Allan McRae <allan@archlinux.org>
Allow finding which mirror was used to fetch a file.
This makes it a bit easier to debug situations in which mirrors serve
bad files with HTTP 200.
Signed-off-by: Vladimir Panteleev <archlinux@cy.md>
When downloading in parallel, sort by package size so that the larger
packages are queued first to fully leverage parallelism.
Addresses FS#70172
Signed-off-by: Charlie Sale <softwaresale01@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
Github and other sites redirect their downloads to a cdn. So the
download http://foo.org/myrepo.db may redirect to something like
https://cdn.foo.org/83749327439.
This then causes pacman to try and download the sig as
https://cdn.foo.org/83749327439.sig which is incorrect. In this case
pacman should append .sig to the original url.
However urls like https://archlinux.org/packages/community/x86_64/0ad/download/
Redirect to the mirror, so .sig has to appended after the redirects and
not before.
So we decide if we should append .sig on the original or effective url
based on if the effective url (minus the query part) has .db or .pkg in it.
Fixes FS#71148
---
v2: move variable decleration to start of block
v3: use dbext instead of db
archweb's download links all ended in /download. This cause all the temp
files to be named download.part. With parallel downloads this results in
multiple downloads to go to the same temp file and breaks the transaction.
Assign random temporary filenames to downloads from URLs that are either
missing a filename, or if the filename does not contain at least three
hyphens (as a well formed package filename does).
While this approach to determining when to use a temporary filename is
not 100% foolproof, it does keep nice looking download progress bar names
when a proper package filename is given. The only downside of not using
temporary files when provided with a filename with three or more hyphens
is URLs created specifically to bypass temporary filename usage can not
be downloaded in parallel. We probably do not want to download packages
from such URLs anyway.
Fixes FS#71464
Modified-by: Allan McRae (do not use temporary files for realish URLs)
Signed-off-by: Allan McRae <allan@archlinux.org>
If the original download redirects to to a different url then alpm would
try to name the sig file after the url instead of <original_file>.sig.
Instead force this naming scheme regardless of url.
Fixes FS#71274
Signed-off-by: Allan McRae <allan@archlinux.org>
Around the same time retry events were added, there was a patch to pass
sig download events to the frontend. The retry code was not updated to
account for this.
Signed-off-by: morganamilo <morganamilo@archlinux.org>
Signed-off-by: Allan McRae <allan@archlinux.org>
Some servers respond with error pages (e.g. 404.html) when a package is
not present. These were getting written to packages before moving onto
the next server. Reset the download progress on 400+ error conditions
to avoid this.
Signed-off-by: Allan McRae <allan@archlinux.org>
When a download fails on one mirror a new download is started on the
next mirror. This causes the ammount downloaded to reset, confusing the
rate math and making it display a negative rate.
This is further complicated by the fact that a download may be resumed
from where it is or started over.
To account for this we alert the frontend that the download was
restarted. Pacman then starts the progress bar over.
Signed-off-by: Allan McRae <allan@archlinux.org>
Restore the prior indicator whether or not databases were up to date.
0 is used to indicate if *any* db was actually updated as callers are
more likely to care about that than if *all* dbs were updated.
Signed-off-by: Andrew Gregory <andrew.gregory.8@gmail.com>
An extra break causes _alpm_download to break out of the payload loop as
soon as it sees a successful url download with XferCommand.
Fixes: FS#70608 - -U fails to download all files with XferCommand
Signed-off-by: Andrew Gregory <andrew.gregory.8@gmail.com>
Our callbacks require front-ends to maintain state in order to provide
reasonable output. The new download callback in particular requires
much more complex state information to be saved. Without the ability to
provide context, state must be saved globally, which may not be possible
for all front-ends. Scripting language bindings in particular have no
way to register per-handle callbacks without some form of context.
Implements: FS#12721
Signed-off-by: Andrew Gregory <andrew.gregory.8@gmail.com>
Keep track of errors from servers so that bad ones can be skipped once
a threshold is reached. Key the error tracking off the hostname because
hosts may serve multiple repos under different url's and errors are
likely to be host-wide.
Implements: FS#29293.
Signed-off-by: Andrew Gregory <andrew.gregory.8@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
Total download callback called right before packages start downloaded.
But we already have an event for such event (ALPM_EVENT_PKG_RETRIEVE_START)
and it is naturally to use the event to pass information about expected
download size.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
In some cases (when trust_remote_name is used for a URL without a filename and
no Content-Disposition is provided by the server) destfile_name will be
NULL. In this case payload data will be stored in tempfile_name and no
destfile_name is set.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
At the end of payload use it calls _alpm_dload_payload_reset()
that will free() these and other fields anyway.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
The main payload final name might be affected by url redirects or
Content-Disposition HTTP header value.
We want to make sure that accompanion *.sig filename always matches the
package filename. So ignore finalname/Content-Disposition for the *.sig file.
It also helps to fix a corner case when the download URL does not contain
a filename and server provides Content-Disposition for the main payload
but not for the signature payload.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
In case if *.pkg exists but *.sig file does not we still have to pass
the pkg to multi_download API.
To avoid redownloading *.pkg file we use CURLOPT_TIMECONDITION curl option.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
Until now callee of ALPM download functionality has been in charge of
payload creation both for the main file (e.g. *.pkg) and for the accompanied
*.sig file. One advantage of such solution is that all payloads are
independent and can be fetched in parallel thus exploiting the maximum
level of download parallelism.
To build *.sig file url we've been using a simple string concatenation:
$requested_url + ".sig". Unfortunately there are cases when it does not
work. For example an archlinux.org "Download From Mirror" link looks like
this https://www.archlinux.org/packages/core/x86_64/bash/download/ and
it gets redirected to some mirror. But if we append ".sig" to the end of
the link url and try to download it then archlinux.org returns 404 error.
To overcome this issue we need to follow redirects for the main payload
first, find the final url and only then append '.sig' suffix.
This implies 2 things:
- the signature payload initialization need to be moved to dload.c
as it is the place where we have access to the resolved url
- *.sig is downloaded serially with the main payload and this reduces
level of parallelism
Move *.sig payload creation to dload.c. Once the main payload is fetched
successfully we check if the callee asked to download the accompanied
signature. If yes - create a new payload and add it to mcurl.
*.sig payload does not use server list of the main payload and thus does
not support mirror failover. *.sig file comes from the same server as
the main payload.
Refactor event loop in curl_multi_download_internal() a bit. Instead of
relying on curl_multi_check_finished_download() to return number of new
payloads we simply rerun the loop iteration one more time to check if
there are any active downloads left.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
All users of _alpm_download() have been refactored to the new API.
It is time to remove the old _alpm_download() functionality now.
This change also removes obsolete SIGPIPE signal handler functionality
(this is a leftover from libfetch days).
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Installing remote packages using its URL is an interesting case for ALPM
API. Unlike package sync ('pacman -S pkg1 pkg2') '-U' does not deal with
server mirror list. Thus _alpm_multi_download() should be able to
handle file download for payloads that either have 'fileurl' field
or pair of fields ('servers' and 'filepath') set.
Signature for alpm_fetch_pkgurl() has changed and it accepts an
output list that is populated with filepaths to fetched packages.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Create a list of dload_payloads and pass it to the new _alpm_multi_*
interface.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
Multiplexed download requires ability to draw UI for multiple active progress
bars. To implement it we use ANSI codes to move cursor up/down and then
redraw the required progress bar.
`pacman_multibar_ui.active_downloads` field represents the list of active
downloads that correspond to progress bars.
`struct pacman_progress_bar` is a data structure for a progress bar.
In some cases (e.g. database downloads) we want to keep progress bars in order.
In some other cases (package downloads) we want to move completed items to the
top of the screen. Function `multibar_move_completed_up` allows to configure
such behavior.
Per discussion in the maillist we do not want to show download progress for
signature files.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
With the previous download interface the callback uses the first progress
event as 'download has started' signal. Unfortunately it does not work with
up-to-date files that never receive 'download progress' events.
Up-to-date database messages are currently handled in sync_syncdbs()
after the sequential download is completed and a result from ALPM is
received. But this is not going to work with multiplexed download
interface that returns the result only after all files are completed.
Another problem with 'first progress event is the beginning of the
download' is that such events time are unpredictable. Thus the UI progress
bar order might differ from what has been passed by client to
alpm_dbs_update() function. We actually want to keep the dbs progress bars
in a strict order.
To help to solve the given problems extend the download callback to
allow 2 more events - download started and completed. 'Download started'
events appear in the same order as in the list given by a client.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
curl_multi_download_internal() is the main loop that creates up to
'ParallelDownloads' easy curl handles, adds them to mcurl and then
performs curl execution. This is when the paralled downloads happens.
Once any of the downloads complete the function checks its result.
In case if the download fails it initiates retry with the next server
from payload->servers list. At the download completion all the payload
resources are cleaned up.
curl_multi_check_finished_download() is essentially refactored version of
curl_download_internal() adopted for multi_curl. Once mcurl porting is
complete curl_download_internal() will be removed.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
It is an equivalent of _alpm_download but accepts a list of payloads.
curl_multi_download_internal() is a stub at this moment and will be
implemented in the later commits of this patch series.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
dload_payload->curlerr is a field that is used inside
curl_download_internal() function only. It can be converted to a local
variable.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
To be able to run multiple download in parallel efficiently we need to
use curl_multi interface [1]. It introduces a set of APIs over new type
of handler 'CURLM'.
Create CURLM object at the application start and set it to global ALPM
context.
The 'single-download' CURL handle moves to payload struct. A new CURL
handle is created for each payload with intention to be processed by CURLM.
Note that curl_download_internal() is not ported to CURLM interface due
to the fact that the function will go away soon.
[1] https://curl.haxx.se/libcurl/c/libcurl-multi.html
Signed-off-by: Allan McRae <allan@archlinux.org>
This is an equivalent of alpm_db_update but for multiplexed (parallel)
download. The difference is that this function accepts list of
databases to update. And then ALPM internals download it in parallel if
possible.
Add a stub for _alpm_multi_download the function that will do parallel
payloads downloads in the future.
Introduce dload_payload->filepath field that contains url path to the
file we download. It is like fileurl field but does not contain
protocol/server part. The rationale for having this field is that with
the curl multidownload the server retry logic is going to move to a curl
callback. And the callback needs to be able to reconstruct the 'next'
fileurl. One will be able to do it by getting the next server url from
'servers' list and then concat with filepath. Once the 'parallel download'
refactoring is over 'fileurl' field will go away.
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
The GOTO_ERR define was added in commit 80ae8014 for use in future commits.
There are plenty of places in the code base it can be used, so convert them.
Signed-off-by: Allan McRae <allan@archlinux.org>
Commit e6a6d307 detected complete part files by comparing a payload's
max_size to initial_size. However, these values are also equal when we
use pacman -U on a URL as max_size is set to 0 in that case. Add a further
condition to avoid that.
Signed-off-by: Allan McRae <allan@archlinux.org>
In rare cases, likely due to a well timed Ctrl+C, but possibly due to a
broken mirror, a ".part" file may have size at least that of the correct
package size.
When encountering this issue, currently pacman fails in different ways
depending on where the package falls in the list to download. If last,
"wrong or NULL argument passed" error is reported, or a "invalid or
corrupt package" issue if not.
Capture these .part files, and remove the extension. This lets pacman
either use the package if valid, or offer to remove it if it fails checksum
or signature verification.
Signed-off-by: Allan McRae <allan@archlinux.org>
