Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 23.01.2020 15:52, Julien Grall wrote: > Therefore, they will have to accept whatever string is reported by > HVMLoader (or Xen). As you already allow Xen to configure it, why would > that be a problem to change the one in Kconfig? Why do you need to fix > it up in hvmloader as well? Because, as stated before, hvmloader is actually the presentation layer from the guest firmware pov. Hence what is sensibly coming back as "" or "" from the hypercall should not propagate into the firmware tables the guest gets to see. Other users of the hypercall may very well leave these strings unfiltered, such that to consumers it's clear what has happened (and from other context it would then typically also be clear what exact piece of information it is which has got hidden). Jan ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 1/23/20 2:52 PM, Julien Grall wrote: > Hi, > > On 23/01/2020 14:45, George Dunlap wrote: >> On 1/23/20 2:42 PM, Julien Grall wrote: >>> Hi, >>> >>> On 23/01/2020 11:32, Sergey Dyasli wrote: On 22/01/2020 11:25, Julien Grall wrote: > > > On 22/01/2020 11:19, Sergey Dyasli wrote: >> On 22/01/2020 10:14, Julien Grall wrote: >>> >>> >>> On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: > On 17.01.2020 17:44, Sergey Dyasli wrote: >> v2 --> v3: >> - Remove hvmloader filtering > > Why? Seeing the prior discussion, how about adding > XENVER_denied to > return the "denied" string, allowing components which want to > filter > to know exactly what to look for? And then re-add the filtering > you > had? (The help text of the config option should then perhaps be > extended to make very clear that the chosen string should not > match > anything that could potentially be returned by any of the XENVER_ > sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. >>> >>> While I am not a maintainer of this code, my concern is you impose >>> the conversion from "denied" to "" to all the users (include those >>> who wants to keep "denied"). >> >> This is not what's happening here: the default is still "" >> (as >> per patch 1); but patch 2 makes XENVER_extraversion, >> XENVER_compile_info >> and XENVER_changeset to return "" instead of the real values >> which causes the UI / logs issues. > > I was referring the SMBios filtering... I don't think doing a blank > filtering is the right thing to do in the hvmloader for the reason > explained above. Apologies for misunderstanding the context. But I disagree about hvmloader. Returning "denied" from xen_version hypercall to guests is one thing, but hvmloader and SMBios tables are parts of the hypervisor and putting "denied" there is simply a terrible user experience. >>> >>> I am not going to comment on the user experience because this is up to >>> another bikeshedding. The question is which string will you use? Why an >>> empty string over something different? >>> >>> However, if an admin has control over the "deny" string, why would he >>> ever want to filter it in hvmloader? Why can't he just replace the one >>> in Kconfig? >> >> Most admins don't compile their own version of Xen... > > Those admins are also unlikely going to build there own hvmloader, right? > > Therefore, they will have to accept whatever string is reported by > HVMLoader (or Xen). As you already allow Xen to configure it, why would > that be a problem to change the one in Kconfig? Why do you need to fix > it up in hvmloader as well? Right, the idea was perhaps as upstream, we should modify hvmloader to *always* replace "" with "". (Or potentially with a more benign string, like "hypervisor build unknown".) -George ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
Hi, On 23/01/2020 14:45, George Dunlap wrote: On 1/23/20 2:42 PM, Julien Grall wrote: Hi, On 23/01/2020 11:32, Sergey Dyasli wrote: On 22/01/2020 11:25, Julien Grall wrote: On 22/01/2020 11:19, Sergey Dyasli wrote: On 22/01/2020 10:14, Julien Grall wrote: On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. While I am not a maintainer of this code, my concern is you impose the conversion from "denied" to "" to all the users (include those who wants to keep "denied"). This is not what's happening here: the default is still "" (as per patch 1); but patch 2 makes XENVER_extraversion, XENVER_compile_info and XENVER_changeset to return "" instead of the real values which causes the UI / logs issues. I was referring the SMBios filtering... I don't think doing a blank filtering is the right thing to do in the hvmloader for the reason explained above. Apologies for misunderstanding the context. But I disagree about hvmloader. Returning "denied" from xen_version hypercall to guests is one thing, but hvmloader and SMBios tables are parts of the hypervisor and putting "denied" there is simply a terrible user experience. I am not going to comment on the user experience because this is up to another bikeshedding. The question is which string will you use? Why an empty string over something different? However, if an admin has control over the "deny" string, why would he ever want to filter it in hvmloader? Why can't he just replace the one in Kconfig? Most admins don't compile their own version of Xen... Those admins are also unlikely going to build there own hvmloader, right? Therefore, they will have to accept whatever string is reported by HVMLoader (or Xen). As you already allow Xen to configure it, why would that be a problem to change the one in Kconfig? Why do you need to fix it up in hvmloader as well? Cheers, -- Julien Grall ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 1/23/20 2:42 PM, Julien Grall wrote: > Hi, > > On 23/01/2020 11:32, Sergey Dyasli wrote: >> On 22/01/2020 11:25, Julien Grall wrote: >>> >>> >>> On 22/01/2020 11:19, Sergey Dyasli wrote: On 22/01/2020 10:14, Julien Grall wrote: > > > On 22/01/2020 10:01, Sergey Dyasli wrote: >> On 20/01/2020 10:01, Jan Beulich wrote: >>> On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering >>> >>> Why? Seeing the prior discussion, how about adding XENVER_denied to >>> return the "denied" string, allowing components which want to filter >>> to know exactly what to look for? And then re-add the filtering you >>> had? (The help text of the config option should then perhaps be >>> extended to make very clear that the chosen string should not match >>> anything that could potentially be returned by any of the XENVER_ >>> sub-ops.) >> >> I had the following reasoning: >> >> 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" >> anyway. >> >> 2. Filtering in DMI tables is not a complete solution, since denied >> string leaks elsewhere through the hypercall (PV guests, sysfs, >> driver >> logs) as Andrew has pointed out in the previous discussion. >> >> On the other hand, SMBios filtering slightly improves the >> situation for >> HVM domains, so I can return it if maintainers find it worthy. > > While I am not a maintainer of this code, my concern is you impose > the conversion from "denied" to "" to all the users (include those > who wants to keep "denied"). This is not what's happening here: the default is still "" (as per patch 1); but patch 2 makes XENVER_extraversion, XENVER_compile_info and XENVER_changeset to return "" instead of the real values which causes the UI / logs issues. >>> >>> I was referring the SMBios filtering... I don't think doing a blank >>> filtering is the right thing to do in the hvmloader for the reason >>> explained above. >> >> Apologies for misunderstanding the context. But I disagree about >> hvmloader. Returning "denied" from xen_version hypercall to guests is >> one thing, but hvmloader and SMBios tables are parts of the hypervisor >> and putting "denied" there is simply a terrible user experience. > > I am not going to comment on the user experience because this is up to > another bikeshedding. The question is which string will you use? Why an > empty string over something different? > > However, if an admin has control over the "deny" string, why would he > ever want to filter it in hvmloader? Why can't he just replace the one > in Kconfig? Most admins don't compile their own version of Xen... -George ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
Hi, On 23/01/2020 11:32, Sergey Dyasli wrote: On 22/01/2020 11:25, Julien Grall wrote: On 22/01/2020 11:19, Sergey Dyasli wrote: On 22/01/2020 10:14, Julien Grall wrote: On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. While I am not a maintainer of this code, my concern is you impose the conversion from "denied" to "" to all the users (include those who wants to keep "denied"). This is not what's happening here: the default is still "" (as per patch 1); but patch 2 makes XENVER_extraversion, XENVER_compile_info and XENVER_changeset to return "" instead of the real values which causes the UI / logs issues. I was referring the SMBios filtering... I don't think doing a blank filtering is the right thing to do in the hvmloader for the reason explained above. Apologies for misunderstanding the context. But I disagree about hvmloader. Returning "denied" from xen_version hypercall to guests is one thing, but hvmloader and SMBios tables are parts of the hypervisor and putting "denied" there is simply a terrible user experience. I am not going to comment on the user experience because this is up to another bikeshedding. The question is which string will you use? Why an empty string over something different? However, if an admin has control over the "deny" string, why would he ever want to filter it in hvmloader? Why can't he just replace the one in Kconfig? Cheers, -- Julien Grall ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 1/22/20 11:44 AM, Sergey Dyasli wrote: > On 22/01/2020 10:57, George Dunlap wrote: >> On 1/22/20 10:14 AM, Julien Grall wrote: >>> >>> >>> On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: > On 17.01.2020 17:44, Sergey Dyasli wrote: >> v2 --> v3: >> - Remove hvmloader filtering > > Why? Seeing the prior discussion, how about adding XENVER_denied to > return the "denied" string, allowing components which want to filter > to know exactly what to look for? And then re-add the filtering you > had? (The help text of the config option should then perhaps be > extended to make very clear that the chosen string should not match > anything that could potentially be returned by any of the XENVER_ > sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. >>> >>> While I am not a maintainer of this code, my concern is you impose the >>> conversion from "denied" to "" to all the users (include those who wants >>> to keep "denied"). >>> >>> If you were doing any filtering in hvmloader, then it would be best if >>> this is configurable. But this is a bit pointless if you already allow >>> the user to configure the string at the hypervisor level :). >> >> So there are two things we're concerned about: >> - Some people don't want to scare users with a "" string >> - Some people don't want to "silently fail" with a "" string >> >> The fact is, in *both cases*, this is a UI problem. EVERY caller of >> this interface should figure out independently what a graceful way of >> handling failure is for their target UI. Any caller who does not think >> carefully about what to do in the failure case is buggy -- which >> includes every single caller today. The CONFIG_XSM_DENIED_STRING is a >> gross hack fallback for buggy UIs. >> >> Now, I don't like to tell other people to do work, and I certainly don't >> plan on fixing hvmloader at the moment, because it's low-priority for >> me. But I do think that having hvmloader detect failure and explicitly >> make a sensible decision is the right thing to do, regardless of the >> availability of CONFIG_XSM_DENIED_STRING to work around buggy callers. > > It's not entirely clear to me what you suggest to do with hvmloader. > Could you elaborate a bit? Well, basically "think carefully about the user experience and decide what to do". Given your comment in response to Julien, I would think that would mean checking for CONFIG_XSM_DENIED_STRING in hvmloader and replacing it with nothing (or some other indication that's more user-friendly). Perhaps re-submitting your hvmloader patch as a follow-up patch. But as I said, it's just a suggestion. -George ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22/01/2020 11:25, Julien Grall wrote: > > > On 22/01/2020 11:19, Sergey Dyasli wrote: >> On 22/01/2020 10:14, Julien Grall wrote: >>> >>> >>> On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: > On 17.01.2020 17:44, Sergey Dyasli wrote: >> v2 --> v3: >> - Remove hvmloader filtering > > Why? Seeing the prior discussion, how about adding XENVER_denied to > return the "denied" string, allowing components which want to filter > to know exactly what to look for? And then re-add the filtering you > had? (The help text of the config option should then perhaps be > extended to make very clear that the chosen string should not match > anything that could potentially be returned by any of the XENVER_ > sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. >>> >>> While I am not a maintainer of this code, my concern is you impose the >>> conversion from "denied" to "" to all the users (include those who wants to >>> keep "denied"). >> >> This is not what's happening here: the default is still "" (as >> per patch 1); but patch 2 makes XENVER_extraversion, XENVER_compile_info >> and XENVER_changeset to return "" instead of the real values >> which causes the UI / logs issues. > > I was referring the SMBios filtering... I don't think doing a blank filtering > is the right thing to do in the hvmloader for the reason explained above. Apologies for misunderstanding the context. But I disagree about hvmloader. Returning "denied" from xen_version hypercall to guests is one thing, but hvmloader and SMBios tables are parts of the hypervisor and putting "denied" there is simply a terrible user experience. > > Regarding CONFIG_XSM_DENIED_STRING, I think this is a good step as it allows > the vendor to configure it. -- Thanks, Sergey ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22/01/2020 12:32, Jan Beulich wrote: On 22.01.2020 13:05, Julien Grall wrote: Hi George, On 22/01/2020 10:57, George Dunlap wrote: On 1/22/20 10:14 AM, Julien Grall wrote: On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. While I am not a maintainer of this code, my concern is you impose the conversion from "denied" to "" to all the users (include those who wants to keep "denied"). If you were doing any filtering in hvmloader, then it would be best if this is configurable. But this is a bit pointless if you already allow the user to configure the string at the hypervisor level :). So there are two things we're concerned about: - Some people don't want to scare users with a "" string - Some people don't want to "silently fail" with a "" string The fact is, in *both cases*, this is a UI problem. EVERY caller of this interface should figure out independently what a graceful way of handling failure is for their target UI. Any caller who does not think carefully about what to do in the failure case is buggy -- which includes every single caller today. The CONFIG_XSM_DENIED_STRING is a gross hack fallback for buggy UIs. I agree that the two cases you explained are UI problems. However, I can see other use for the Kconfig option (with some tweaks). At AWS, consistency accross two stable versions is very important. So most of the version strings exposed to the guest are fixed. Therefore a guest can be migrated seemlessly between two different versions without seen any change that may break it. A guest aware of being run on a hypervisor would also be aware that it may be migrated, and hence that the version of the underlying hypervisor may change (if it cares about versions in the first place). If you use upstream-as-is yes. But with the on-going discussion regarding live udpate and guest transparent migration, a guest would seemlessly move between Xen versions without even been aware. A guest unaware of being run on a hypervisor wouldn't care about version and alike strings at all. Nevertheless I'm sure you play this game for a (real world) reason, e.g. people making wrong assumptions. But is this something you really think the upstream hypervisor should be made care about? I agree that upstream does not necessarily needs it today. But this is an example on how configurable version strings could be useful by downstream users. Cheers, -- Julien Grall ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22.01.2020 13:05, Julien Grall wrote: > Hi George, > > On 22/01/2020 10:57, George Dunlap wrote: >> On 1/22/20 10:14 AM, Julien Grall wrote: >>> >>> >>> On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: > On 17.01.2020 17:44, Sergey Dyasli wrote: >> v2 --> v3: >> - Remove hvmloader filtering > > Why? Seeing the prior discussion, how about adding XENVER_denied to > return the "denied" string, allowing components which want to filter > to know exactly what to look for? And then re-add the filtering you > had? (The help text of the config option should then perhaps be > extended to make very clear that the chosen string should not match > anything that could potentially be returned by any of the XENVER_ > sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. >>> >>> While I am not a maintainer of this code, my concern is you impose the >>> conversion from "denied" to "" to all the users (include those who wants >>> to keep "denied"). >>> >>> If you were doing any filtering in hvmloader, then it would be best if >>> this is configurable. But this is a bit pointless if you already allow >>> the user to configure the string at the hypervisor level :). >> >> So there are two things we're concerned about: >> - Some people don't want to scare users with a "" string >> - Some people don't want to "silently fail" with a "" string >> >> The fact is, in *both cases*, this is a UI problem. EVERY caller of >> this interface should figure out independently what a graceful way of >> handling failure is for their target UI. Any caller who does not think >> carefully about what to do in the failure case is buggy -- which >> includes every single caller today. The CONFIG_XSM_DENIED_STRING is a >> gross hack fallback for buggy UIs. > > I agree that the two cases you explained are UI problems. However, I can > see other use for the Kconfig option (with some tweaks). > > At AWS, consistency accross two stable versions is very important. So > most of the version strings exposed to the guest are fixed. Therefore a > guest can be migrated seemlessly between two different versions without > seen any change that may break it. A guest aware of being run on a hypervisor would also be aware that it may be migrated, and hence that the version of the underlying hypervisor may change (if it cares about versions in the first place). A guest unaware of being run on a hypervisor wouldn't care about version and alike strings at all. Nevertheless I'm sure you play this game for a (real world) reason, e.g. people making wrong assumptions. But is this something you really think the upstream hypervisor should be made care about? Jan ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
Hi George, On 22/01/2020 10:57, George Dunlap wrote: On 1/22/20 10:14 AM, Julien Grall wrote: On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. While I am not a maintainer of this code, my concern is you impose the conversion from "denied" to "" to all the users (include those who wants to keep "denied"). If you were doing any filtering in hvmloader, then it would be best if this is configurable. But this is a bit pointless if you already allow the user to configure the string at the hypervisor level :). So there are two things we're concerned about: - Some people don't want to scare users with a "" string - Some people don't want to "silently fail" with a "" string The fact is, in *both cases*, this is a UI problem. EVERY caller of this interface should figure out independently what a graceful way of handling failure is for their target UI. Any caller who does not think carefully about what to do in the failure case is buggy -- which includes every single caller today. The CONFIG_XSM_DENIED_STRING is a gross hack fallback for buggy UIs. I agree that the two cases you explained are UI problems. However, I can see other use for the Kconfig option (with some tweaks). At AWS, consistency accross two stable versions is very important. So most of the version strings exposed to the guest are fixed. Therefore a guest can be migrated seemlessly between two different versions without seen any change that may break it. You could imagine using XSM to know whether you want to expose the guest the real or fixed version strings. Put it that way, this sounds less a gross hack over "". The use case described would require multiple Kconfig options though. Now, I don't like to tell other people to do work, and I certainly don't plan on fixing hvmloader at the moment, because it's low-priority for me. But I do think that having hvmloader detect failure and explicitly make a sensible decision is the right thing to do, regardless of the availability of CONFIG_XSM_DENIED_STRING to work around buggy callers. The lengthy discussion about returning "" shows that we (XenProject community) are not in position to decide what is the sensible value here (even for filtering). By allowing a vendor to configure the string in the hypervisor, you remove that decision from us. Cheers, -- Julien Grall ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22/01/2020 10:57, George Dunlap wrote: > On 1/22/20 10:14 AM, Julien Grall wrote: >> >> >> On 22/01/2020 10:01, Sergey Dyasli wrote: >>> On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: > v2 --> v3: > - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) >>> >>> I had the following reasoning: >>> >>> 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. >>> >>> 2. Filtering in DMI tables is not a complete solution, since denied >>> string leaks elsewhere through the hypercall (PV guests, sysfs, driver >>> logs) as Andrew has pointed out in the previous discussion. >>> >>> On the other hand, SMBios filtering slightly improves the situation for >>> HVM domains, so I can return it if maintainers find it worthy. >> >> While I am not a maintainer of this code, my concern is you impose the >> conversion from "denied" to "" to all the users (include those who wants >> to keep "denied"). >> >> If you were doing any filtering in hvmloader, then it would be best if >> this is configurable. But this is a bit pointless if you already allow >> the user to configure the string at the hypervisor level :). > > So there are two things we're concerned about: > - Some people don't want to scare users with a "" string > - Some people don't want to "silently fail" with a "" string > > The fact is, in *both cases*, this is a UI problem. EVERY caller of > this interface should figure out independently what a graceful way of > handling failure is for their target UI. Any caller who does not think > carefully about what to do in the failure case is buggy -- which > includes every single caller today. The CONFIG_XSM_DENIED_STRING is a > gross hack fallback for buggy UIs. > > Now, I don't like to tell other people to do work, and I certainly don't > plan on fixing hvmloader at the moment, because it's low-priority for > me. But I do think that having hvmloader detect failure and explicitly > make a sensible decision is the right thing to do, regardless of the > availability of CONFIG_XSM_DENIED_STRING to work around buggy callers. It's not entirely clear to me what you suggest to do with hvmloader. Could you elaborate a bit? -- Thanks, Sergey ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22/01/2020 11:19, Sergey Dyasli wrote: On 22/01/2020 10:14, Julien Grall wrote: On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. While I am not a maintainer of this code, my concern is you impose the conversion from "denied" to "" to all the users (include those who wants to keep "denied"). This is not what's happening here: the default is still "" (as per patch 1); but patch 2 makes XENVER_extraversion, XENVER_compile_info and XENVER_changeset to return "" instead of the real values which causes the UI / logs issues. I was referring the SMBios filtering... I don't think doing a blank filtering is the right thing to do in the hvmloader for the reason explained above. Regarding CONFIG_XSM_DENIED_STRING, I think this is a good step as it allows the vendor to configure it. Cheers, -- Julien Grall ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22/01/2020 10:14, Julien Grall wrote: > > > On 22/01/2020 10:01, Sergey Dyasli wrote: >> On 20/01/2020 10:01, Jan Beulich wrote: >>> On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering >>> >>> Why? Seeing the prior discussion, how about adding XENVER_denied to >>> return the "denied" string, allowing components which want to filter >>> to know exactly what to look for? And then re-add the filtering you >>> had? (The help text of the config option should then perhaps be >>> extended to make very clear that the chosen string should not match >>> anything that could potentially be returned by any of the XENVER_ >>> sub-ops.) >> >> I had the following reasoning: >> >> 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. >> >> 2. Filtering in DMI tables is not a complete solution, since denied >> string leaks elsewhere through the hypercall (PV guests, sysfs, driver >> logs) as Andrew has pointed out in the previous discussion. >> >> On the other hand, SMBios filtering slightly improves the situation for >> HVM domains, so I can return it if maintainers find it worthy. > > While I am not a maintainer of this code, my concern is you impose the > conversion from "denied" to "" to all the users (include those who wants to > keep "denied"). This is not what's happening here: the default is still "" (as per patch 1); but patch 2 makes XENVER_extraversion, XENVER_compile_info and XENVER_changeset to return "" instead of the real values which causes the UI / logs issues. > > If you were doing any filtering in hvmloader, then it would be best if this > is configurable. But this is a bit pointless if you already allow the user to > configure the string at the hypervisor level :). -- Thanks, Sergey ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 1/22/20 10:14 AM, Julien Grall wrote: > > > On 22/01/2020 10:01, Sergey Dyasli wrote: >> On 20/01/2020 10:01, Jan Beulich wrote: >>> On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering >>> >>> Why? Seeing the prior discussion, how about adding XENVER_denied to >>> return the "denied" string, allowing components which want to filter >>> to know exactly what to look for? And then re-add the filtering you >>> had? (The help text of the config option should then perhaps be >>> extended to make very clear that the chosen string should not match >>> anything that could potentially be returned by any of the XENVER_ >>> sub-ops.) >> >> I had the following reasoning: >> >> 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. >> >> 2. Filtering in DMI tables is not a complete solution, since denied >> string leaks elsewhere through the hypercall (PV guests, sysfs, driver >> logs) as Andrew has pointed out in the previous discussion. >> >> On the other hand, SMBios filtering slightly improves the situation for >> HVM domains, so I can return it if maintainers find it worthy. > > While I am not a maintainer of this code, my concern is you impose the > conversion from "denied" to "" to all the users (include those who wants > to keep "denied"). > > If you were doing any filtering in hvmloader, then it would be best if > this is configurable. But this is a bit pointless if you already allow > the user to configure the string at the hypervisor level :). So there are two things we're concerned about: - Some people don't want to scare users with a "" string - Some people don't want to "silently fail" with a "" string The fact is, in *both cases*, this is a UI problem. EVERY caller of this interface should figure out independently what a graceful way of handling failure is for their target UI. Any caller who does not think carefully about what to do in the failure case is buggy -- which includes every single caller today. The CONFIG_XSM_DENIED_STRING is a gross hack fallback for buggy UIs. Now, I don't like to tell other people to do work, and I certainly don't plan on fixing hvmloader at the moment, because it's low-priority for me. But I do think that having hvmloader detect failure and explicitly make a sensible decision is the right thing to do, regardless of the availability of CONFIG_XSM_DENIED_STRING to work around buggy callers. -George ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22/01/2020 10:01, Sergey Dyasli wrote: On 20/01/2020 10:01, Jan Beulich wrote: On 17.01.2020 17:44, Sergey Dyasli wrote: v2 --> v3: - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. While I am not a maintainer of this code, my concern is you impose the conversion from "denied" to "" to all the users (include those who wants to keep "denied"). If you were doing any filtering in hvmloader, then it would be best if this is configurable. But this is a bit pointless if you already allow the user to configure the string at the hypervisor level :). Cheers, -- Julien Grall ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 22.01.2020 11:01, Sergey Dyasli wrote: > On 20/01/2020 10:01, Jan Beulich wrote: >> On 17.01.2020 17:44, Sergey Dyasli wrote: >>> v2 --> v3: >>> - Remove hvmloader filtering >> >> Why? Seeing the prior discussion, how about adding XENVER_denied to >> return the "denied" string, allowing components which want to filter >> to know exactly what to look for? And then re-add the filtering you >> had? (The help text of the config option should then perhaps be >> extended to make very clear that the chosen string should not match >> anything that could potentially be returned by any of the XENVER_ >> sub-ops.) > > I had the following reasoning: > > 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. > > 2. Filtering in DMI tables is not a complete solution, since denied > string leaks elsewhere through the hypercall (PV guests, sysfs, driver > logs) as Andrew has pointed out in the previous discussion. Well, that's Andrew's thinking. To me, getting back "" from the hypercall is not a leaking of this string, but the intended output. I continue to think that getting back a blank string there is confusing, as it is far more likely to be mistaken to be actually blank than it is for "" to be mistaken for actual data. Where such a string wants filtering is to be determined for every consumer of this interface separately; as said before, this is a UI function, not something to be catered for in the hypervisor. Jan > On the other hand, SMBios filtering slightly improves the situation for > HVM domains, so I can return it if maintainers find it worthy. > > -- > Thanks, > Sergey > ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 20/01/2020 10:01, Jan Beulich wrote: > On 17.01.2020 17:44, Sergey Dyasli wrote: >> v2 --> v3: >> - Remove hvmloader filtering > > Why? Seeing the prior discussion, how about adding XENVER_denied to > return the "denied" string, allowing components which want to filter > to know exactly what to look for? And then re-add the filtering you > had? (The help text of the config option should then perhaps be > extended to make very clear that the chosen string should not match > anything that could potentially be returned by any of the XENVER_ > sub-ops.) I had the following reasoning: 1. Most real-world users would set CONFIG_XSM_DENIED_STRING="" anyway. 2. Filtering in DMI tables is not a complete solution, since denied string leaks elsewhere through the hypercall (PV guests, sysfs, driver logs) as Andrew has pointed out in the previous discussion. On the other hand, SMBios filtering slightly improves the situation for HVM domains, so I can return it if maintainers find it worthy. -- Thanks, Sergey ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v3 2/2] xsm: hide detailed Xen version from unprivileged guests
On 17.01.2020 17:44, Sergey Dyasli wrote: > v2 --> v3: > - Remove hvmloader filtering Why? Seeing the prior discussion, how about adding XENVER_denied to return the "denied" string, allowing components which want to filter to know exactly what to look for? And then re-add the filtering you had? (The help text of the config option should then perhaps be extended to make very clear that the chosen string should not match anything that could potentially be returned by any of the XENVER_ sub-ops.) Jan ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel