all lists on lists.proxmox.com
 help / color / mirror / Atom feed
From: Fiona Ebner <f.ebner@proxmox.com>
To: "Dominik Csapak" <d.csapak@proxmox.com>,
	"Proxmox VE development discussion" <pve-devel@lists.proxmox.com>,
	"Fabian Grünbichler" <f.gruenbichler@proxmox.com>
Subject: Re: [pve-devel] [PATCH v2 qemu-server 1/1] disk import: add additional safeguards for imported image files
Date: Fri, 15 Nov 2024 11:05:19 +0100	[thread overview]
Message-ID: <0a2c9e96-ca76-42e3-9165-4bb4e7871bf7@proxmox.com> (raw)
In-Reply-To: <787aaaa9-1f7a-49c8-925e-49308d24df0d@proxmox.com>

On 15.11.24 10:55 AM, Dominik Csapak wrote:
> On 11/15/24 10:49, Fiona Ebner wrote:
>> On 15.11.24 10:42 AM, Fiona Ebner wrote:
>>> On 04.11.24 11:42 AM, Fabian Grünbichler wrote:
>>>> creating non-raw disk images with arbitrary content is only possible
>>>> with raw
>>>> access to the storage, but checking for references to external files
>>>> doesn't
>>>> hurt.
>>>>
>>>> Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
>>>> ---
>>>>
>>>> Notes:
>>>>      requires pve-storage change to actually have an effect
>>>>           v2:
>>>>      - re-order code to improve readability
>>>>      - extract $path into variable to avoid scalar vs array context
>>>> issue
>>>>
>>>>   PVE/API2/Qemu.pm | 18 +++++++++++++++---
>>>>   1 file changed, 15 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/PVE/API2/Qemu.pm b/PVE/API2/Qemu.pm
>>>> index 848001b6..ae0c39bf 100644
>>>> --- a/PVE/API2/Qemu.pm
>>>> +++ b/PVE/API2/Qemu.pm
>>>> @@ -412,18 +412,29 @@ my sub create_disks : prototype($$$$$$$$$$) {
>>>>             $needs_creation = $live_import;
>>>>   -        if (PVE::Storage::parse_volume_id($source, 1)) { # PVE-
>>>> managed volume
>>>> +        my ($source_storage, $source_volid) =
>>>> PVE::Storage::parse_volume_id($source, 1);
>>>> +
>>>> +        if ($source_storage) { # PVE-managed volume
>>>>               if ($live_import && $ds ne 'efidisk0') {
>>>>               my $path = PVE::Storage::path($storecfg, $source)
>>>>                   or die "failed to get a path for '$source'\n";
>>>>               $source = $path;
>>>> -            ($size, my $source_format) =
>>>> PVE::Storage::file_size_info($source);
>>>> +            # check potentially untrusted image file!
>>>> +            ($size, my $source_format) =
>>>> PVE::Storage::file_size_info($source, undef, 1);
>>>> +
>>>>               die "could not get file size of $source\n" if !$size;
>>>>               $live_import_mapping->{$ds} = {
>>>>                   path => $source,
>>>>                   format => $source_format,
>>>>               };
>>>>               } else {
>>>> +            # check potentially untrusted image file!
>>>> +            my $scfg = PVE::Storage::storage_config($storecfg,
>>>> $source_storage);
>>>> +            if ($scfg->{path}) {
>>>
>>> Is there a special reason this is made conditional on the presence of
>>> $scfg->{path}? Note that the above check for live import isn't.
>>> Shouldn't matter much right now (except if there is a weird third-party
>>> plugin that has qcow2 and doesn't use $scfg->{path}), but as long as we
>>> get a result for PVE::Storage::path(), we should be able to call
>>> file_size_info() too, right?
>>>
>>
>> Or maybe not, but then live-import should've used volume_size_info() too
>> I guess? I think it would be good to add an 'untrusted' flag to
>> volume_size_info() too and have the storage layer do the right thing,
>> rather than manually checking $scfg->{path} here.
>>
>>
> my 2 cents:
> 
> the checking of untrusted images is mainly helpful for file based volumes
> (such as qcow2/vmdk/etc.) where we can detect backing images etc.
> 

Yes, currently. But like this, third-party plugins without a path won't
be able to have checks and maybe we'd like to add different checks in
the future that won't require file-based volumes.

My point is about the mismatch between the check here and the
live-import one. Either we can call file_size_info() for any result of
path() (not sure, because e.g. ISCSIDirectPlugin's path() will return an
'iscsi://' protocol path for exmplae) or we should use
volume_size_info() instead.

But I guess that's orthogonal and can still be improved later.

> the check for 'path' is how we ususually determine if a storage is
> file-based or not (would probably better to have that check
> in storage and it should probably check for qcow2/vmdk/etc. support)
> 
> but we already use that check in some places
> 


_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

  reply	other threads:[~2024-11-15 10:05 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-11-04 10:42 [pve-devel] [PATCH v2 storage/guest-common/qemu-server 0/3] harden import of file-based volumes Fabian Grünbichler
2024-11-04 10:42 ` [pve-devel] [PATCH v2 guest-common 1/1] storage tunnel: check just-imported image files Fabian Grünbichler
2025-04-03 10:01   ` [pve-devel] applied: " Thomas Lamprecht
2024-11-04 10:42 ` [pve-devel] [PATCH v2 storage 1/1] file_size_info: implement untrusted mode Fabian Grünbichler
2024-11-07 12:16   ` Fiona Ebner
2024-11-14  9:33   ` Dominik Csapak
2024-11-14 18:14   ` [pve-devel] applied: " Thomas Lamprecht
2024-11-04 10:42 ` [pve-devel] [PATCH v2 qemu-server 1/1] disk import: add additional safeguards for imported image files Fabian Grünbichler
2024-11-14  9:34   ` Dominik Csapak
2024-11-15  9:42   ` Fiona Ebner
2024-11-15  9:49     ` Fiona Ebner
2024-11-15  9:55       ` Dominik Csapak
2024-11-15 10:05         ` Fiona Ebner [this message]
2024-11-15 10:16           ` Dominik Csapak
2024-11-15 10:15   ` Fiona Ebner

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=0a2c9e96-ca76-42e3-9165-4bb4e7871bf7@proxmox.com \
    --to=f.ebner@proxmox.com \
    --cc=d.csapak@proxmox.com \
    --cc=f.gruenbichler@proxmox.com \
    --cc=pve-devel@lists.proxmox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal