Message ID | 20200116170004.14373-1-jgross@suse.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | xen/balloon: Support xend-based toolstack take two | expand |
On 1/16/20 12:00 PM, Juergen Gross wrote: > Commit 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") > tried to fix a regression with running on rather ancient Xen versions. > Unfortunately the fix was based on the assumption that xend would > just use another Xenstore node, but in reality only some downstream > versions of xend are doing that. The upstream xend does not write > that Xenstore node at all, so the problem must be fixed in another > way. > > The easiest way to achieve that is to fall back to the behavior before > commit 5266b8e4445c ("xen: fix booting ballooned down hvm guest") > in case the static memory maximum can't be read. > > Fixes: 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") > Signed-off-by: Juergen Gross <jgross@suse.com> > Cc: <stable@vger.kernel.org> # 4.13 Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
On 16.01.2020 18:00, Juergen Gross wrote: > Commit 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") > tried to fix a regression with running on rather ancient Xen versions. > Unfortunately the fix was based on the assumption that xend would > just use another Xenstore node, but in reality only some downstream > versions of xend are doing that. The upstream xend does not write > that Xenstore node at all, so the problem must be fixed in another > way. > > The easiest way to achieve that is to fall back to the behavior before > commit 5266b8e4445c ("xen: fix booting ballooned down hvm guest") > in case the static memory maximum can't be read. I could use some help here: Prior to said commit there was target_diff = new_target - balloon_stats.target_pages; Which is, afaict, ... > --- a/drivers/xen/xen-balloon.c > +++ b/drivers/xen/xen-balloon.c > @@ -94,7 +94,7 @@ static void watch_target(struct xenbus_watch *watch, > "%llu", &static_max) == 1)) > static_max >>= PAGE_SHIFT - 10; > else > - static_max = new_target; > + static_max = balloon_stats.current_pages; > > target_diff = (xen_pv_domain() || xen_initial_domain()) ? 0 > : static_max - balloon_stats.target_pages; ... what the code does before your change. Afaict there was never a use of balloon_stats.current_pages in this function. Jan
On 17.01.20 12:01, Jan Beulich wrote: > On 16.01.2020 18:00, Juergen Gross wrote: >> Commit 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") >> tried to fix a regression with running on rather ancient Xen versions. >> Unfortunately the fix was based on the assumption that xend would >> just use another Xenstore node, but in reality only some downstream >> versions of xend are doing that. The upstream xend does not write >> that Xenstore node at all, so the problem must be fixed in another >> way. >> >> The easiest way to achieve that is to fall back to the behavior before >> commit 5266b8e4445c ("xen: fix booting ballooned down hvm guest") >> in case the static memory maximum can't be read. > > I could use some help here: Prior to said commit there was > > target_diff = new_target - balloon_stats.target_pages; > > > Which is, afaict, ... > >> --- a/drivers/xen/xen-balloon.c >> +++ b/drivers/xen/xen-balloon.c >> @@ -94,7 +94,7 @@ static void watch_target(struct xenbus_watch *watch, >> "%llu", &static_max) == 1)) >> static_max >>= PAGE_SHIFT - 10; >> else >> - static_max = new_target; >> + static_max = balloon_stats.current_pages; >> >> target_diff = (xen_pv_domain() || xen_initial_domain()) ? 0 >> : static_max - balloon_stats.target_pages; > > ... what the code does before your change. Afaict there was > never a use of balloon_stats.current_pages in this function. That is a little bit indirect, yes. In the end I want static_max to be either the maximum reported by Xen, or if not available, the current assumed memory size, which can be found in balloon_stats.current_pages. The main idea is to avoid a negative target_diff which would result in not ballooning down. Juergen
On 17.01.2020 12:31, Jürgen Groß wrote: > On 17.01.20 12:01, Jan Beulich wrote: >> On 16.01.2020 18:00, Juergen Gross wrote: >>> Commit 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") >>> tried to fix a regression with running on rather ancient Xen versions. >>> Unfortunately the fix was based on the assumption that xend would >>> just use another Xenstore node, but in reality only some downstream >>> versions of xend are doing that. The upstream xend does not write >>> that Xenstore node at all, so the problem must be fixed in another >>> way. >>> >>> The easiest way to achieve that is to fall back to the behavior before >>> commit 5266b8e4445c ("xen: fix booting ballooned down hvm guest") >>> in case the static memory maximum can't be read. >> >> I could use some help here: Prior to said commit there was >> >> target_diff = new_target - balloon_stats.target_pages; >> >> >> Which is, afaict, ... >> >>> --- a/drivers/xen/xen-balloon.c >>> +++ b/drivers/xen/xen-balloon.c >>> @@ -94,7 +94,7 @@ static void watch_target(struct xenbus_watch *watch, >>> "%llu", &static_max) == 1)) >>> static_max >>= PAGE_SHIFT - 10; >>> else >>> - static_max = new_target; >>> + static_max = balloon_stats.current_pages; >>> >>> target_diff = (xen_pv_domain() || xen_initial_domain()) ? 0 >>> : static_max - balloon_stats.target_pages; >> >> ... what the code does before your change. Afaict there was >> never a use of balloon_stats.current_pages in this function. > > That is a little bit indirect, yes. In the end I want static_max to > be either the maximum reported by Xen, or if not available, the current > assumed memory size, which can be found in balloon_stats.current_pages. > > The main idea is to avoid a negative target_diff which would result in > not ballooning down. All understood. Yet the change is then not a restore of prior behavior (just in a limited case), but a change to behavior that we never there before. I.e. it was indeed my assumption that the code was right, but the description was misleading. Jan
On 17.01.20 12:36, Jan Beulich wrote: > On 17.01.2020 12:31, Jürgen Groß wrote: >> On 17.01.20 12:01, Jan Beulich wrote: >>> On 16.01.2020 18:00, Juergen Gross wrote: >>>> Commit 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") >>>> tried to fix a regression with running on rather ancient Xen versions. >>>> Unfortunately the fix was based on the assumption that xend would >>>> just use another Xenstore node, but in reality only some downstream >>>> versions of xend are doing that. The upstream xend does not write >>>> that Xenstore node at all, so the problem must be fixed in another >>>> way. >>>> >>>> The easiest way to achieve that is to fall back to the behavior before >>>> commit 5266b8e4445c ("xen: fix booting ballooned down hvm guest") >>>> in case the static memory maximum can't be read. >>> >>> I could use some help here: Prior to said commit there was >>> >>> target_diff = new_target - balloon_stats.target_pages; >>> >>> >>> Which is, afaict, ... >>> >>>> --- a/drivers/xen/xen-balloon.c >>>> +++ b/drivers/xen/xen-balloon.c >>>> @@ -94,7 +94,7 @@ static void watch_target(struct xenbus_watch *watch, >>>> "%llu", &static_max) == 1)) >>>> static_max >>= PAGE_SHIFT - 10; >>>> else >>>> - static_max = new_target; >>>> + static_max = balloon_stats.current_pages; >>>> >>>> target_diff = (xen_pv_domain() || xen_initial_domain()) ? 0 >>>> : static_max - balloon_stats.target_pages; >>> >>> ... what the code does before your change. Afaict there was >>> never a use of balloon_stats.current_pages in this function. >> >> That is a little bit indirect, yes. In the end I want static_max to >> be either the maximum reported by Xen, or if not available, the current >> assumed memory size, which can be found in balloon_stats.current_pages. >> >> The main idea is to avoid a negative target_diff which would result in >> not ballooning down. > > All understood. Yet the change is then not a restore of prior behavior > (just in a limited case), but a change to behavior that we never there > before. I.e. it was indeed my assumption that the code was right, but > the description was misleading. The description is misleading as it fails to mention commit 96edd61dcf44362d3e, which introduced target_diff. I'll add that to the commit message. Juergen
diff --git a/drivers/xen/xen-balloon.c b/drivers/xen/xen-balloon.c index 6d12fc368210..a8d24433c8e9 100644 --- a/drivers/xen/xen-balloon.c +++ b/drivers/xen/xen-balloon.c @@ -94,7 +94,7 @@ static void watch_target(struct xenbus_watch *watch, "%llu", &static_max) == 1)) static_max >>= PAGE_SHIFT - 10; else - static_max = new_target; + static_max = balloon_stats.current_pages; target_diff = (xen_pv_domain() || xen_initial_domain()) ? 0 : static_max - balloon_stats.target_pages;
Commit 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") tried to fix a regression with running on rather ancient Xen versions. Unfortunately the fix was based on the assumption that xend would just use another Xenstore node, but in reality only some downstream versions of xend are doing that. The upstream xend does not write that Xenstore node at all, so the problem must be fixed in another way. The easiest way to achieve that is to fall back to the behavior before commit 5266b8e4445c ("xen: fix booting ballooned down hvm guest") in case the static memory maximum can't be read. Fixes: 3aa6c19d2f38be ("xen/balloon: Support xend-based toolstack") Signed-off-by: Juergen Gross <jgross@suse.com> Cc: <stable@vger.kernel.org> # 4.13 --- drivers/xen/xen-balloon.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)