[pbs-devel] [PATCH v3 proxmox-backup 12/20] file-restore-daemon: add disk module

[Stefan Reiter <s.reiter@proxmox.com>]

Includes functionality for scanning and referring to partitions on
attached disks (i.e. snapshot images).

Fairly modular structure, so adding ZFS/LVM/etc... support in the future
should be easy.

The path is encoded as "/disk/bucket/component/path/to/file", e.g.
"/drive-scsi0/part/0/etc/passwd". See the comments for further
explanations on the design.

Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
---

v3:
* fix ZFS exclusion in scan(), style cleanups

 src/bin/proxmox-restore-daemon.rs      |  16 +-
 src/bin/proxmox_restore_daemon/disk.rs | 329 +++++++++++++++++++++++++
 src/bin/proxmox_restore_daemon/mod.rs  |   3 +
 3 files changed, 347 insertions(+), 1 deletion(-)
 create mode 100644 src/bin/proxmox_restore_daemon/disk.rs

diff --git a/src/bin/proxmox-restore-daemon.rs b/src/bin/proxmox-restore-daemon.rs
index 6b453ad3..a2701b7c 100644
--- a/src/bin/proxmox-restore-daemon.rs
+++ b/src/bin/proxmox-restore-daemon.rs
@@ -1,13 +1,14 @@
 ///! Daemon binary to run inside a micro-VM for secure single file restore of disk images
 use anyhow::{bail, format_err, Error};
 use log::error;
+use lazy_static::lazy_static;
 
 use std::os::unix::{
     io::{FromRawFd, RawFd},
     net,
 };
 use std::path::Path;
-use std::sync::Arc;
+use std::sync::{Arc, Mutex};
 
 use tokio::sync::mpsc;
 use tokio_stream::wrappers::ReceiverStream;
@@ -26,6 +27,13 @@ pub const MAX_PENDING: usize = 32;
 /// Will be present in base initramfs
 pub const VM_DETECT_FILE: &str = "/restore-vm-marker";
 
+lazy_static! {
+    /// The current disks state. Use for accessing data on the attached snapshots.
+    pub static ref DISK_STATE: Arc<Mutex<DiskState>> = {
+        Arc::new(Mutex::new(DiskState::scan().unwrap()))
+    };
+}
+
 /// This is expected to be run by 'proxmox-file-restore' within a mini-VM
 fn main() -> Result<(), Error> {
     if !Path::new(VM_DETECT_FILE).exists() {
@@ -41,6 +49,12 @@ fn main() -> Result<(), Error> {
         .write_style(env_logger::WriteStyle::Never)
         .init();
 
+    // scan all attached disks now, before starting the API
+    // this will panic and stop the VM if anything goes wrong
+    {
+        let _disk_state = DISK_STATE.lock().unwrap();
+    }
+
     proxmox_backup::tools::runtime::main(run())
 }
 
diff --git a/src/bin/proxmox_restore_daemon/disk.rs b/src/bin/proxmox_restore_daemon/disk.rs
new file mode 100644
index 00000000..f9d7c8aa
--- /dev/null
+++ b/src/bin/proxmox_restore_daemon/disk.rs
@@ -0,0 +1,329 @@
+//! Low-level disk (image) access functions for file restore VMs.
+use anyhow::{bail, format_err, Error};
+use lazy_static::lazy_static;
+use log::{info, warn};
+
+use std::collections::HashMap;
+use std::fs::{create_dir_all, File};
+use std::io::{BufRead, BufReader};
+use std::path::{Component, Path, PathBuf};
+
+use proxmox::const_regex;
+use proxmox::tools::fs;
+use proxmox_backup::api2::types::BLOCKDEVICE_NAME_REGEX;
+
+const_regex! {
+    VIRTIO_PART_REGEX = r"^vd[a-z]+(\d+)$";
+}
+
+lazy_static! {
+    static ref FS_OPT_MAP: HashMap<&'static str, &'static str> = {
+        let mut m = HashMap::new();
+
+        // otherwise ext complains about mounting read-only
+        m.insert("ext2", "noload");
+        m.insert("ext3", "noload");
+        m.insert("ext4", "noload");
+
+        // ufs2 is used as default since FreeBSD 5.0 released in 2003, so let's assume that
+        // whatever the user is trying to restore is not using anything older...
+        m.insert("ufs", "ufstype=ufs2");
+
+        m
+    };
+}
+
+pub enum ResolveResult {
+    Path(PathBuf),
+    BucketTypes(Vec<&'static str>),
+    BucketComponents(Vec<String>),
+}
+
+struct PartitionBucketData {
+    dev_node: String,
+    number: i32,
+    mountpoint: Option<PathBuf>,
+}
+
+/// A "Bucket" represents a mapping found on a disk, e.g. a partition, a zfs dataset or an LV. A
+/// uniquely identifying path to a file then consists of four components:
+/// "/disk/bucket/component/path"
+/// where
+///   disk: fidx file name
+///   bucket: bucket type
+///   component: identifier of the specific bucket
+///   path: relative path of the file on the filesystem indicated by the other parts, may contain
+///         more subdirectories
+/// e.g.: "/drive-scsi0/part/0/etc/passwd"
+enum Bucket {
+    Partition(PartitionBucketData),
+}
+
+impl Bucket {
+    fn filter_mut<'a, A: AsRef<str>, B: AsRef<str>>(
+        haystack: &'a mut Vec<Bucket>,
+        ty: A,
+        comp: B,
+    ) -> Option<&'a mut Bucket> {
+        let ty = ty.as_ref();
+        let comp = comp.as_ref();
+        haystack.iter_mut().find(|b| match b {
+            Bucket::Partition(data) => ty == "part" && comp.parse::<i32>().unwrap() == data.number,
+        })
+    }
+
+    fn type_string(&self) -> &'static str {
+        match self {
+            Bucket::Partition(_) => "part",
+        }
+    }
+
+    fn component_string(&self) -> String {
+        match self {
+            Bucket::Partition(data) => data.number.to_string(),
+        }
+    }
+}
+
+/// Functions related to the local filesystem. This mostly exists so we can use 'supported_fs' in
+/// try_mount while a Bucket is still mutably borrowed from DiskState.
+struct Filesystems {
+    supported_fs: Vec<String>,
+}
+
+impl Filesystems {
+    fn scan() -> Result<Self, Error> {
+        // detect kernel supported filesystems
+        let mut supported_fs = Vec::new();
+        for f in BufReader::new(File::open("/proc/filesystems")?)
+            .lines()
+            .filter_map(Result::ok)
+        {
+            // ZFS is treated specially, don't attempt to do a regular mount with it
+            let f = f.trim();
+            if !f.starts_with("nodev") && f != "zfs" {
+                supported_fs.push(f.to_owned());
+            }
+        }
+
+        Ok(Self { supported_fs })
+    }
+
+    fn ensure_mounted(&self, bucket: &mut Bucket) -> Result<PathBuf, Error> {
+        match bucket {
+            Bucket::Partition(data) => {
+                // regular data partition à la "/dev/vdxN"
+                if let Some(mp) = &data.mountpoint {
+                    return Ok(mp.clone());
+                }
+
+                let mp = format!("/mnt{}/", data.dev_node);
+                self.try_mount(&data.dev_node, &mp)?;
+                let mp = PathBuf::from(mp);
+                data.mountpoint = Some(mp.clone());
+                Ok(mp)
+            }
+        }
+    }
+
+    fn try_mount(&self, source: &str, target: &str) -> Result<(), Error> {
+        use nix::mount::*;
+
+        create_dir_all(target)?;
+
+        // try all supported fs until one works - this is the way Busybox's 'mount' does it too:
+        // https://git.busybox.net/busybox/tree/util-linux/mount.c?id=808d93c0eca49e0b22056e23d965f0d967433fbb#n2152
+        // note that ZFS is intentionally left out (see scan())
+        let flags =
+            MsFlags::MS_RDONLY | MsFlags::MS_NOEXEC | MsFlags::MS_NOSUID | MsFlags::MS_NODEV;
+        for fs in &self.supported_fs {
+            let fs: &str = fs.as_ref();
+            let opts = FS_OPT_MAP.get(fs).copied();
+            match mount(Some(source), target, Some(fs), flags, opts) {
+                Ok(()) => {
+                    info!("mounting '{}' succeeded, fstype: '{}'", source, fs);
+                    return Ok(());
+                }
+                Err(err) => {
+                    warn!("mount error on '{}' ({}) - {}", source, fs, err);
+                }
+            }
+        }
+
+        bail!("all mounts failed or no supported file system")
+    }
+}
+
+pub struct DiskState {
+    filesystems: Filesystems,
+    disk_map: HashMap<String, Vec<Bucket>>,
+}
+
+impl DiskState {
+    /// Scan all disks for supported buckets.
+    pub fn scan() -> Result<Self, Error> {
+        // create mapping for virtio drives and .fidx files (via serial description)
+        // note: disks::DiskManager relies on udev, which we don't have
+        let mut disk_map = HashMap::new();
+        for entry in proxmox_backup::tools::fs::scan_subdir(
+            libc::AT_FDCWD,
+            "/sys/block",
+            &BLOCKDEVICE_NAME_REGEX,
+        )?
+        .filter_map(Result::ok)
+        {
+            let name = unsafe { entry.file_name_utf8_unchecked() };
+            if !name.starts_with("vd") {
+                continue;
+            }
+
+            let sys_path: &str = &format!("/sys/block/{}", name);
+
+            let serial = fs::file_read_string(&format!("{}/serial", sys_path));
+            let fidx = match serial {
+                Ok(serial) => serial,
+                Err(err) => {
+                    warn!("disk '{}': could not read serial file - {}", name, err);
+                    continue;
+                }
+            };
+
+            let mut parts = Vec::new();
+            for entry in proxmox_backup::tools::fs::scan_subdir(
+                libc::AT_FDCWD,
+                sys_path,
+                &VIRTIO_PART_REGEX,
+            )?
+            .filter_map(Result::ok)
+            {
+                let part_name = unsafe { entry.file_name_utf8_unchecked() };
+                let devnode = format!("/dev/{}", part_name);
+                let part_path = format!("/sys/block/{}/{}", name, part_name);
+
+                // create partition device node for further use
+                let dev_num_str = fs::file_read_firstline(&format!("{}/dev", part_path))?;
+                let (major, minor) = dev_num_str.split_at(dev_num_str.find(':').unwrap());
+                Self::mknod_blk(&devnode, major.parse()?, minor[1..].trim_end().parse()?)?;
+
+                let number = fs::file_read_firstline(&format!("{}/partition", part_path))?
+                    .trim()
+                    .parse::<i32>()?;
+
+                info!(
+                    "drive '{}' ('{}'): found partition '{}' ({})",
+                    name, fidx, devnode, number
+                );
+
+                let bucket = Bucket::Partition(PartitionBucketData {
+                    dev_node: devnode,
+                    mountpoint: None,
+                    number,
+                });
+
+                parts.push(bucket);
+            }
+
+            disk_map.insert(fidx.to_owned(), parts);
+        }
+
+        Ok(Self {
+            filesystems: Filesystems::scan()?,
+            disk_map,
+        })
+    }
+
+    /// Given a path like "/drive-scsi0.img.fidx/part/0/etc/passwd", this will mount the first
+    /// partition of 'drive-scsi0' on-demand (i.e. if not already mounted) and return a path
+    /// pointing to the requested file locally, e.g. "/mnt/vda1/etc/passwd", which can be used to
+    /// read the file.  Given a partial path, i.e. only "/drive-scsi0.img.fidx" or
+    /// "/drive-scsi0.img.fidx/part", it will return a list of available bucket types or bucket
+    /// components respectively
+    pub fn resolve(&mut self, path: &Path) -> Result<ResolveResult, Error> {
+        let mut cmp = path.components().peekable();
+        match cmp.peek() {
+            Some(Component::RootDir) | Some(Component::CurDir) => {
+                cmp.next();
+            }
+            None => bail!("empty path cannot be resolved to file location"),
+            _ => {}
+        }
+
+        let req_fidx = match cmp.next() {
+            Some(Component::Normal(x)) => x.to_string_lossy(),
+            _ => bail!("no or invalid image in path"),
+        };
+
+        let buckets = match self.disk_map.get_mut(req_fidx.as_ref()) {
+            Some(x) => x,
+            None => bail!("given image '{}' not found", req_fidx),
+        };
+
+        let bucket_type = match cmp.next() {
+            Some(Component::Normal(x)) => x.to_string_lossy(),
+            Some(c) => bail!("invalid bucket in path: {:?}", c),
+            None => {
+                // list bucket types available
+                let mut types = buckets
+                    .iter()
+                    .map(|b| b.type_string())
+                    .collect::<Vec<&'static str>>();
+                // dedup requires duplicates to be consecutive, which is the case - see scan()
+                types.dedup();
+                return Ok(ResolveResult::BucketTypes(types));
+            }
+        };
+
+        let component = match cmp.next() {
+            Some(Component::Normal(x)) => x.to_string_lossy(),
+            Some(c) => bail!("invalid bucket component in path: {:?}", c),
+            None => {
+                // list bucket components available
+                let comps = buckets
+                    .iter()
+                    .filter(|b| b.type_string() == bucket_type)
+                    .map(Bucket::component_string)
+                    .collect();
+                return Ok(ResolveResult::BucketComponents(comps));
+            }
+        };
+
+        let mut bucket = match Bucket::filter_mut(buckets, &bucket_type, &component) {
+            Some(bucket) => bucket,
+            None => bail!(
+                "bucket/component path not found: {}/{}/{}",
+                req_fidx,
+                bucket_type,
+                component
+            ),
+        };
+
+        // bucket found, check mount
+        let mountpoint = self
+            .filesystems
+            .ensure_mounted(&mut bucket)
+            .map_err(|err| {
+                format_err!(
+                    "mounting '{}/{}/{}' failed: {}",
+                    req_fidx,
+                    bucket_type,
+                    component,
+                    err
+                )
+            })?;
+
+        let mut local_path = PathBuf::new();
+        local_path.push(mountpoint);
+        for rem in cmp {
+            local_path.push(rem);
+        }
+
+        Ok(ResolveResult::Path(local_path))
+    }
+
+    fn mknod_blk(path: &str, maj: u64, min: u64) -> Result<(), Error> {
+        use nix::sys::stat;
+        let dev = stat::makedev(maj, min);
+        stat::mknod(path, stat::SFlag::S_IFBLK, stat::Mode::S_IRWXU, dev)?;
+        Ok(())
+    }
+}
diff --git a/src/bin/proxmox_restore_daemon/mod.rs b/src/bin/proxmox_restore_daemon/mod.rs
index 3b52cf06..58e2bb6e 100644
--- a/src/bin/proxmox_restore_daemon/mod.rs
+++ b/src/bin/proxmox_restore_daemon/mod.rs
@@ -6,3 +6,6 @@ pub mod auth;
 
 mod watchdog;
 pub use watchdog::*;
+
+mod disk;
+pub use disk::*;
-- 
2.20.1