On Tue, 23 Oct 2012 13:40:13 +0800 Lei Li <li...@linux.vnet.ibm.com> wrote:
> >> +static int cirmem_chr_write(CharDriverState *chr, const uint8_t *buf, int > >> len) > >> +{ > >> + CirMemCharDriver *d = chr->opaque; > >> + int i; > >> + > >> + if (len < 0) { > >> + return -1; > >> + } > >> + > >> + /* The size should be a power of 2. */ > > Shouldn't you enforce that, then? > > Yes, it has been checked when open the CirMemChar backend in patch 2/5, > as code below: > > if (d->size & (d->size -1)) { > return NULL; > } You could add an assert() in cirmem_chr_write() then.