Searching for an unused bounce buffer in mReservedMemBitmap and
reserving the buffer by flipping the bit is a critical section
which must not be interrupted. Raise the TPL level to ensure
that.
Without this fix it can happen that IoMmuDxe hands out the same
bounce buffer twice, causing trouble do
On 19/07/2023 12:33, Gerd Hoffmann wrote:
Searching for an unused bounce buffer in mReservedMemBitmap and
reserving the buffer by flipping the bit is a critical section
which must not be interrupted. Raise the TPL level to ensure
that.
Without this fix it can happen that IoMmuDxe hands out the
On Wed, Jul 19, 2023 at 04:04:28PM +, Michael Brown wrote:
> On 19/07/2023 12:33, Gerd Hoffmann wrote:
> > Searching for an unused bounce buffer in mReservedMemBitmap and
> > reserving the buffer by flipping the bit is a critical section
> > which must not be interrupted. Raise the TPL level t
On Wed, 19 Jul 2023 at 18:32, Gerd Hoffmann wrote:
>
> On Wed, Jul 19, 2023 at 04:04:28PM +, Michael Brown wrote:
> > On 19/07/2023 12:33, Gerd Hoffmann wrote:
> > > Searching for an unused bounce buffer in mReservedMemBitmap and
> > > reserving the buffer by flipping the bit is a critical sec
On 19/07/2023 17:52, Ard Biesheuvel wrote:
On Wed, 19 Jul 2023 at 18:32, Gerd Hoffmann wrote:
On Wed, Jul 19, 2023 at 04:04:28PM +, Michael Brown wrote:
It looks as though IoMmuFreeBounceBuffer() should also raise to TPL_NOTIFY
while modifying mReservedMemBitmap, since the modification mad
On Wed, 19 Jul 2023 at 19:40, Michael Brown wrote:
>
> On 19/07/2023 17:52, Ard Biesheuvel wrote:
> > On Wed, 19 Jul 2023 at 18:32, Gerd Hoffmann wrote:
> >> On Wed, Jul 19, 2023 at 04:04:28PM +, Michael Brown wrote:
> >>> It looks as though IoMmuFreeBounceBuffer() should also raise to TPL_NO
> > > mReservedMemBitmap &= (UINT32)(~MapInfo->ReservedMemBitmap);
> >
> > I'd expect modern compilers optimize that to a single instruction,
>
> You mean something along the lines of
>
> andl %reg, mReservedMemBitmap(%rip)
>
> right?
Yes.
> > but
> > yes, it's not guaranteed to happen, th
Hi,
> > On a quick review of the code, there appear to be other points that also
> > modify mReservedMemBitmap (IoMmuAllocateCommonBuffer() and
> > IoMmuFreeCommonBuffer()). I'd guess that these also need to raise to
> > TPL_NOTIFY, but I'm not familiar with the code so I don't know if
> > ther
On Thu, 20 Jul 2023 at 10:28, Gerd Hoffmann wrote:
>
> > > > mReservedMemBitmap &= (UINT32)(~MapInfo->ReservedMemBitmap);
> > >
> > > I'd expect modern compilers optimize that to a single instruction,
> >
> > You mean something along the lines of
> >
> > andl %reg, mReservedMemBitmap(%rip)
> >
