On Mon, Jul 07, 2014 at 02:18:07PM -0400, John Snow wrote: > +/*** IO macros for the AHCI memory registers. ***/ > +#define void_incr(vptr, OFST) ((void *)((char *)(vptr) + (OFST)))
I'm pretty sure QEMU takes advantage of GCC's void pointer arithmetic extension: https://gcc.gnu.org/onlinedocs/gcc-4.9.1/gcc/Pointer-Arith.html#Pointer-Arith In other words, vptr + OFST works and you don't need a macro. > +#define ahci_set(regno, mask) ahci_wreg((regno), ahci_rreg(regno) | (mask)) > +#define ahci_clr(regno, mask) ahci_wreg((regno), ahci_rreg(regno) & ~(mask)) Unused. Please move to the patch that actually uses them. > +#define px_set(port, reg, mask) px_wreg((port), (reg), \ > + px_rreg((port), (reg)) | (mask)); > +#define px_clr(port, reg, mask) px_wreg((port), (reg), \ > + px_rreg((port), (reg)) & ~(mask)); Unused. Please move to the patch that actually uses them. > + /* We need to know the size of the region, > + * but qpci_iomap doesn't save it. Recalculate it. */ It seems like many tests will want to check the BAR size. Please add an argument to qpci_iomap() so the caller gets the size. > + if (bitset(cap, AHCI_CAP_SAM)) { > + g_test_message("Supports AHCI-Only Mode: GHC_AE is Read-Only."); > + assert_bit_set(reg, AHCI_GHC_AE); > + } else { > + g_test_message("Supports AHCI/Legacy mix."); > + assert_bit_clear(reg, AHCI_GHC_AE); > + } Let's just assert what QEMU implements. > + /* 12 -- 23: Reserved */ > + g_test_message("Verifying HBA reserved area is empty."); Debugging message that can be removed? More elsewhere in this patch.
pgpcudaV9S8K_.pgp
Description: PGP signature
