Message ID | 1563466463-26012-1-git-send-email-wrfsh@yandex-team.ru |
---|---|
State | New |
Headers | show |
Series | [v2] i386/acpi: fix gint overflow in crs_range_compare | expand |
On Thu, Jul 18, 2019 at 07:14:23PM +0300, Evgeny Yakovlev wrote: > When very large regions (32GB sized in our case, PCI pass-through of GPUs) > are compared substraction result does not fit into gint. > > As a result crs_replace_with_free_ranges does not get sorted ranges and > incorrectly computes PCI64 free space regions. Which then makes linux > guest complain about device and PCI64 hole intersection and device > becomes unusable. > > Fix that by returning exactly fitting ranges. > > Also fix indentation of an entire crs_replace_with_free_ranges to make > checkpatch happy. > > Signed-off-by: Evgeny Yakovlev <wrfsh@yandex-team.ru> queued, thanks a lot! > --- > v2: > entire crs_replace_with_free_ranges was indented with 5 spaces, including my change. > fix that as well > > hw/i386/acpi-build.c | 12 +++++++++--- > 1 file changed, 9 insertions(+), 3 deletions(-) > > diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c > index d281ffa..e7b756b 100644 > --- a/hw/i386/acpi-build.c > +++ b/hw/i386/acpi-build.c > @@ -755,10 +755,16 @@ static void crs_range_set_free(CrsRangeSet *range_set) > > static gint crs_range_compare(gconstpointer a, gconstpointer b) > { > - CrsRangeEntry *entry_a = *(CrsRangeEntry **)a; > - CrsRangeEntry *entry_b = *(CrsRangeEntry **)b; > + CrsRangeEntry *entry_a = *(CrsRangeEntry **)a; > + CrsRangeEntry *entry_b = *(CrsRangeEntry **)b; > > - return (int64_t)entry_a->base - (int64_t)entry_b->base; > + if (entry_a->base < entry_b->base) { > + return -1; > + } else if (entry_a->base > entry_b->base) { > + return 1; > + } else { > + return 0; > + } > } > > /* > -- > 2.7.4
diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c index d281ffa..e7b756b 100644 --- a/hw/i386/acpi-build.c +++ b/hw/i386/acpi-build.c @@ -755,10 +755,16 @@ static void crs_range_set_free(CrsRangeSet *range_set) static gint crs_range_compare(gconstpointer a, gconstpointer b) { - CrsRangeEntry *entry_a = *(CrsRangeEntry **)a; - CrsRangeEntry *entry_b = *(CrsRangeEntry **)b; + CrsRangeEntry *entry_a = *(CrsRangeEntry **)a; + CrsRangeEntry *entry_b = *(CrsRangeEntry **)b; - return (int64_t)entry_a->base - (int64_t)entry_b->base; + if (entry_a->base < entry_b->base) { + return -1; + } else if (entry_a->base > entry_b->base) { + return 1; + } else { + return 0; + } } /*
When very large regions (32GB sized in our case, PCI pass-through of GPUs) are compared substraction result does not fit into gint. As a result crs_replace_with_free_ranges does not get sorted ranges and incorrectly computes PCI64 free space regions. Which then makes linux guest complain about device and PCI64 hole intersection and device becomes unusable. Fix that by returning exactly fitting ranges. Also fix indentation of an entire crs_replace_with_free_ranges to make checkpatch happy. Signed-off-by: Evgeny Yakovlev <wrfsh@yandex-team.ru> --- v2: entire crs_replace_with_free_ranges was indented with 5 spaces, including my change. fix that as well hw/i386/acpi-build.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-)