From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from firstgate.proxmox.com (firstgate.proxmox.com [212.224.123.68]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits)) (No client certificate requested) by lists.proxmox.com (Postfix) with ESMTPS id 1A3EF8A99 for ; Tue, 22 Aug 2023 15:05:13 +0200 (CEST) Received: from firstgate.proxmox.com (localhost [127.0.0.1]) by firstgate.proxmox.com (Proxmox) with ESMTP id EC126D2C5 for ; Tue, 22 Aug 2023 15:04:42 +0200 (CEST) Received: from proxmox-new.maurer-it.com (proxmox-new.maurer-it.com [94.136.29.106]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits)) (No client certificate requested) by firstgate.proxmox.com (Proxmox) with ESMTPS for ; Tue, 22 Aug 2023 15:04:41 +0200 (CEST) Received: from proxmox-new.maurer-it.com (localhost.localdomain [127.0.0.1]) by proxmox-new.maurer-it.com (Proxmox) with ESMTP id 3A0F843308 for ; Tue, 22 Aug 2023 15:04:41 +0200 (CEST) Date: Tue, 22 Aug 2023 15:04:39 +0200 From: Wolfgang Bumiller To: Gabriel Goller Cc: pbs-devel@lists.proxmox.com Message-ID: References: <20230821130826.147473-1-g.goller@proxmox.com> <20230821130826.147473-2-g.goller@proxmox.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20230821130826.147473-2-g.goller@proxmox.com> X-SPAM-LEVEL: Spam detection results: 0 AWL 0.103 Adjusted score from AWL reputation of From: address BAYES_00 -1.9 Bayes spam probability is 0 to 1% DMARC_MISSING 0.1 Missing DMARC policy KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record SPF_PASS -0.001 SPF: sender matches SPF record URIBL_BLOCKED 0.001 ADMINISTRATOR NOTICE: The query to URIBL was blocked. See http://wiki.apache.org/spamassassin/DnsBlocklists#dnsbl-block for more information. [extract.rs, create.rs, catalog.rs] Subject: Re: [pbs-devel] [PATCH proxmox-backup v6] fix #4380: check if file is excluded before running `stat()` X-BeenThere: pbs-devel@lists.proxmox.com X-Mailman-Version: 2.1.29 Precedence: list List-Id: Proxmox Backup Server development discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 22 Aug 2023 13:05:13 -0000 On Mon, Aug 21, 2023 at 03:08:26PM +0200, Gabriel Goller wrote: > Passed a closure with the `stat()` function call to `matches()`. This > will traverse through all patterns and try to match using the path only, if a > `file_mode` is needed, it will run the closure. This means that if we exclude > a file with the `MatchType::ANY_FILE_TYPE`, we will skip it without running > `stat()` on it. As we updated the `matches()` function, we also updated all the > invocations of it. > Added `pathpatterns` crate to local overrides in cargo.toml. > > Signed-off-by: Gabriel Goller > --- > > changes v5: > - updated all invocations of `matches()` > > changes v4: > - match only by path and exclude the matched files, the run `stat()` and > match again, this time using the `file_mode`. This will match everything > twice in the worst case, which is not optimal. > changes v3: > - checking for `read` and `execute` permissions before entering directory, > doesn't work because there are a lot of side-effects (executed by > different user, AppArmor, SELinux, ...). > changes v2: > - checking for excluded files with `matches()` before executing `stat()`, > this doesn't work because we get the file_mode from `stat()` and don't > want to ignore it when matching. > > > Cargo.toml | 5 +++-- > pbs-client/src/catalog_shell.rs | 8 +++---- > pbs-client/src/pxar/create.rs | 38 ++++++++++++++++++++------------- > pbs-client/src/pxar/extract.rs | 10 +++++---- > pbs-datastore/src/catalog.rs | 6 +++--- > 5 files changed, 39 insertions(+), 28 deletions(-) > > diff --git a/Cargo.toml b/Cargo.toml > index 5cbae1b8..560794a4 100644 > --- a/Cargo.toml > +++ b/Cargo.toml > @@ -264,8 +264,9 @@ proxmox-rrd.workspace = true > #proxmox-sortable-macro = { path = "../proxmox/proxmox-sortable-macro" } > #proxmox-human-byte = { path = "../proxmox/proxmox-human-byte" } > > -#proxmox-apt = { path = "../proxmox/proxmox-apt" } > -#proxmox-openid = { path = "../proxmox/proxmox-openid" } > +#proxmox-apt = { path = "../proxmox-apt" } > +#proxmox-openid = { path = "../proxmox-openid-rs" } > +#pathpatterns = {path = "../pathpatterns" } > > #pxar = { path = "../pxar" } > > diff --git a/pbs-client/src/catalog_shell.rs b/pbs-client/src/catalog_shell.rs > index b8aaf8cb..f53b3cc5 100644 > --- a/pbs-client/src/catalog_shell.rs > +++ b/pbs-client/src/catalog_shell.rs > @@ -1138,14 +1138,14 @@ impl<'a> ExtractorState<'a> { > pub async fn handle_entry(&mut self, entry: catalog::DirEntry) -> Result<(), Error> { > let match_result = self.match_list.matches(&self.path, entry.get_file_mode()); > let did_match = match match_result { > - Some(MatchType::Include) => true, > - Some(MatchType::Exclude) => false, > - None => self.matches, > + Ok(Some(MatchType::Include)) => true, > + Ok(Some(MatchType::Exclude)) => false, > + _ => self.matches, > }; > > match (did_match, &entry.attr) { > (_, DirEntryAttribute::Directory { .. }) => { > - self.handle_new_directory(entry, match_result).await?; > + self.handle_new_directory(entry, match_result?).await?; > } > (true, DirEntryAttribute::File { .. }) => { > self.dir_stack.push(PathStackEntry::new(entry)); > diff --git a/pbs-client/src/pxar/create.rs b/pbs-client/src/pxar/create.rs > index 2577cf98..2d516cfa 100644 > --- a/pbs-client/src/pxar/create.rs > +++ b/pbs-client/src/pxar/create.rs > @@ -21,7 +21,6 @@ use pxar::Metadata; > > use proxmox_io::vec; > use proxmox_lang::c_str; > -use proxmox_sys::error::SysError; > use proxmox_sys::fs::{self, acl, xattr}; > > use pbs_datastore::catalog::BackupCatalogWriter; > @@ -420,7 +419,7 @@ impl Archiver { > for file in dir.iter() { > let file = file?; > > - let file_name = file.file_name().to_owned(); > + let file_name = file.file_name(); > let file_name_bytes = file_name.to_bytes(); > if file_name_bytes == b"." || file_name_bytes == b".." { > continue; > @@ -434,25 +433,34 @@ impl Archiver { > assert_single_path_component(os_file_name)?; > let full_path = self.path.join(os_file_name); > > - let stat = match nix::sys::stat::fstatat( > + let match_path = PathBuf::from("/").join(full_path.clone()); > + > + let mut stat_results: Option = None; > + > + let get_file_mode = || match nix::sys::stat::fstatat( You don't need that 'match' here if your cases are basically a no-op. You *could* attach the `stat failed on ...` context with the file name here... but this will make it a bit more tedious to check for `ENOENT` as we'd need to use `.downcast()` on the anyhow::Error. So either that, or we'll need to duplicate the context line down below where we do the final `let stat = results.unwrap_or_else(get_file_mode)?`. > dir_fd, > - file_name.as_c_str(), > + file_name.to_owned().as_c_str(), > nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW, > ) { > - Ok(stat) => stat, > - Err(ref err) if err.not_found() => continue, Because the thing is, we still need to handle this case - just noticed that this was removed entirely, which is of course not what we want :-) If the file gets removed between listing it from the directory and us trying to `stat` it, we should just act as if it had never existed. > - Err(err) => return Err(err).context(format!("stat failed on {:?}", full_path)), > + Ok(stat) => Ok(stat), > + Err(e) => Err(e), > }; > - > - let match_path = PathBuf::from("/").join(full_path.clone()); > - if self > - .patterns > - .matches(match_path.as_os_str().as_bytes(), Some(stat.st_mode)) > - == Some(MatchType::Exclude) > + if Some(MatchType::Exclude) > + == self > + .patterns > + .matches(match_path.as_os_str().as_bytes(), || { > + Ok::<_, Errno>(match &stat_results { > + Some(result) => result.st_mode, > + None => stat_results.insert(get_file_mode()?).st_mode, > + }) > + }) ... it will instead need to be checked here ^ Basically, we will need to swap the `if` for a `match` after all, with the cases Ok(Some(MatchType::Exclude)) => continue, Ok(_) => (), Err(err) if err.not_found() => continue, err => return err.with_context(...), or the final 2 cases when using *not* not changing the `get_file_mode` error to `anyhow::Error` would be: Err(err) => match err.downcast::() { Some(err) if err.not_found() => continue, _ => return Err(err), } I hope I didn't miss anything now. > + .with_context(|| format!("stat failed on {full_path:?}"))? > { > continue; > } > > + let stat = stat_results.map(Ok).unwrap_or_else(get_file_mode)?; If the context line is not already in the closure, we'd need to duplicate it here. > + > self.entry_counter += 1; > if self.entry_counter > self.entry_limit { > bail!( > @@ -462,7 +470,7 @@ impl Archiver { > } > > file_list.push(FileListEntry { > - name: file_name, > + name: file_name.to_owned(), > path: full_path, > stat, > }); > @@ -533,7 +541,7 @@ impl Archiver { > let match_path = PathBuf::from("/").join(self.path.clone()); > if self > .patterns > - .matches(match_path.as_os_str().as_bytes(), Some(stat.st_mode)) > + .matches(match_path.as_os_str().as_bytes(), stat.st_mode)? > == Some(MatchType::Exclude) > { > return Ok(()); > diff --git a/pbs-client/src/pxar/extract.rs b/pbs-client/src/pxar/extract.rs > index 4eb6fb90..e24a1560 100644 > --- a/pbs-client/src/pxar/extract.rs > +++ b/pbs-client/src/pxar/extract.rs > @@ -251,22 +251,24 @@ where > > self.extractor.set_path(entry.path().as_os_str().to_owned()); > > + // We can `unwrap()` safely here because we get a `Result<_, std::convert::Infallible>` > let match_result = self.match_list.matches( > entry.path().as_os_str().as_bytes(), > - Some(metadata.file_type() as u32), > - ); > + metadata.file_type() as u32 > + ).unwrap(); > > let did_match = match match_result { > Some(MatchType::Include) => true, > Some(MatchType::Exclude) => false, > - None => self.state.current_match, > + _ => self.state.current_match, > }; > > let extract_res = match (did_match, entry.kind()) { > (_, EntryKind::Directory) => { > self.callback(entry.path()); > > - let create = self.state.current_match && match_result != Some(MatchType::Exclude); > + let create = > + self.state.current_match && match_result != Some(MatchType::Exclude); > let res = self > .extractor > .enter_directory(file_name_os.to_owned(), metadata.clone(), create) > diff --git a/pbs-datastore/src/catalog.rs b/pbs-datastore/src/catalog.rs > index 11c14b64..86e20c92 100644 > --- a/pbs-datastore/src/catalog.rs > +++ b/pbs-datastore/src/catalog.rs > @@ -678,9 +678,9 @@ impl CatalogReader { > } > file_path.extend(&e.name); > match match_list.matches(&file_path, e.get_file_mode()) { > - Some(MatchType::Exclude) => continue, > - Some(MatchType::Include) => callback(file_path)?, > - None => (), > + Ok(Some(MatchType::Exclude)) => continue, > + Ok(Some(MatchType::Include)) => callback(file_path)?, > + _ => (),