Re: [Xen-devel] [PATCH v4 1/2] xsm: add Kconfig option for denied string
On 12/02/2020 09:32, Jan Beulich wrote: > On 11.02.2020 14:42, Sergey Dyasli wrote: >> --- a/xen/common/Kconfig >> +++ b/xen/common/Kconfig >> @@ -228,6 +228,14 @@ choice >> bool "SILO" if XSM_SILO >> endchoice >> >> +config XSM_DENIED_STRING >> +string "xen_version hypercall denied information replacement string" >> +default "" >> +depends on XSM > > Why would this string want to be configurable only in XSM- > enabled builds? For some reason I assumed that xsm_xen_version() is a no-op when CONFIG_XSM is undefined. I can now see that it doesn't depend on any config in which case the dependency (and #ifdef) should indeed be removed. -- Thanks, Sergey ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v4 1/2] xsm: add Kconfig option for denied string
On 11.02.2020 14:56, Andrew Cooper wrote: > On 11/02/2020 13:42, Sergey Dyasli wrote: >> Add Kconfig option to make it possible to configure the string returned >> to non-privileged guests instead of the default "" which could >> propagate to UI / logs after the subsequent patch that hides detailed >> Xen version information from unprivileged guests. >> >> Introduce XENVER_denied_string to allow guests to set up UI / logs >> filtering which dependens on the new CONFIG_XSM_DENIED_STRING. > > No. This is even worse than other suggestions. > > It is entirely unacceptable to expect guests to have to modify them to > figure out when they're being lied to. Why "lied to"? Denying data access is different from lying imo. Plus your proposal to return empty strings then is even more of a lie, for being not even recognizable a "access denied". > And it is now possible *without source code modifications* to create a > Xen which reports one string in this hypercall, and has empty strings > elsewhere, which is even more chaotic for guests. Empty strings elsewhere? Do you mean because of access having been denied, or because they in fact are empty? In the former case I'd like to ask for at least one example, while in the latter case I don't see what wrong you see. Jan ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v4 1/2] xsm: add Kconfig option for denied string
On 11.02.2020 14:42, Sergey Dyasli wrote: > --- a/xen/common/Kconfig > +++ b/xen/common/Kconfig > @@ -228,6 +228,14 @@ choice > bool "SILO" if XSM_SILO > endchoice > > +config XSM_DENIED_STRING > + string "xen_version hypercall denied information replacement string" > + default "" > + depends on XSM Why would this string want to be configurable only in XSM- enabled builds? Jan ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v4 1/2] xsm: add Kconfig option for denied string
On 11/02/2020 13:42, Sergey Dyasli wrote: > Add Kconfig option to make it possible to configure the string returned > to non-privileged guests instead of the default "" which could > propagate to UI / logs after the subsequent patch that hides detailed > Xen version information from unprivileged guests. > > Introduce XENVER_denied_string to allow guests to set up UI / logs > filtering which dependens on the new CONFIG_XSM_DENIED_STRING. No. This is even worse than other suggestions. It is entirely unacceptable to expect guests to have to modify them to figure out when they're being lied to. And it is now possible *without source code modifications* to create a Xen which reports one string in this hypercall, and has empty strings elsewhere, which is even more chaotic for guests. ~Andrew ___ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel