Helge reported that the introduction of PP_MAGIC_MASK let to crashes on boot on his 32-bit parisc machine. The cause of this is the mask is set too wide, so the page_pool_page_is_pp() incurs false positives which crashes the machine. Just disabling the check in page_pool_is_pp() will lead to the page_pool code itself malfunctioning; so instead of doing this, this patch changes the define for PP_DMA_INDEX_BITS to avoid mistaking arbitrary kernel pointers for page_pool-tagged pages. The fix relies on the kernel pointers that alias with the pp_magic field always being above PAGE_OFFSET. With this assumption, we can use the lowest bit of the value of PAGE_OFFSET as the upper bound of the PP_DMA_INDEX_MASK, which should avoid the false positives. Because we cannot rely on PAGE_OFFSET always being a compile-time constant, nor on it always being >0, we fall back to disabling the dma_index storage when there are not enough bits available. This leaves us in the situation we were in before the patch in the Fixes tag, but only on a subset of architecture configurations. This seems to be the best we can do until the transition to page types in complete for page_pool pages. v2: - Make sure there's at least 8 bits available and that the PAGE_OFFSET bit calculation doesn't wrap
Showing the changes to include/linux (1 of 2 files). Show all
Unified · Split
| @@ -4159,14 +4159,13 @@ int arch_lock_shadow_stack_status(struct task_struct *t, unsigned long status); · 4158 unmodified lines | ||
| 4159 | 4159 | * since this value becomes part of PP_SIGNATURE; meaning we can just use the |
| 4160 | 4160 | * space between the PP_SIGNATURE value (without POISON_POINTER_DELTA), and the |
| 4161 | 4161 | * lowest bits of POISON_POINTER_DELTA. On arches where POISON_POINTER_DELTA is |
| 4162 | − * 0, we make sure that we leave the two topmost bits empty, as that guarantees | |
| 4163 | − * we won't mistake a valid kernel pointer for a value we set, regardless of the | |
| 4164 | − * VMSPLIT setting. | |
| 4162 | + * 0, we use the lowest bit of PAGE_OFFSET as the boundary if that value is | |
| 4163 | + * known at compile-time. | |
| 4165 | 4164 | * |
| 4166 | − * Altogether, this means that the number of bits available is constrained by | |
| 4167 | − * the size of an unsigned long (at the upper end, subtracting two bits per the | |
| 4168 | − * above), and the definition of PP_SIGNATURE (with or without | |
| 4169 | − * POISON_POINTER_DELTA). | |
| 4165 | + * If the value of PAGE_OFFSET is not known at compile time, or if it is too | |
| 4166 | + * small to leave at least 8 bits available above PP_SIGNATURE, we define the | |
| 4167 | + * number of bits to be 0, which turns off the DMA index tracking altogether | |
| 4168 | + * (see page_pool_register_dma_index()). | |
| 4170 | 4169 | */ |
| 4171 | 4170 | #define PP_DMA_INDEX_SHIFT (1 + __fls(PP_SIGNATURE - POISON_POINTER_DELTA)) |
| 4172 | 4171 | #if POISON_POINTER_DELTA > 0 |
| @@ -4175,8 +4174,13 @@ int arch_lock_shadow_stack_status(struct task_struct *t, unsigned long status); · 2 unmodified lines | ||
| 4175 | 4174 | */ |
| 4176 | 4175 | #define PP_DMA_INDEX_BITS MIN(32, __ffs(POISON_POINTER_DELTA) - PP_DMA_INDEX_SHIFT) |
| 4177 | 4176 | #else |
| 4178 | −/* Always leave out the topmost two; see above. */ | |
| 4179 | −#define PP_DMA_INDEX_BITS MIN(32, BITS_PER_LONG - PP_DMA_INDEX_SHIFT - 2) | |
| 4177 | +/* Use the lowest bit of PAGE_OFFSET if there's at least 8 bits available; see above */ | |
| 4178 | +#define PP_DMA_INDEX_MIN_OFFSET (1 << (PP_DMA_INDEX_SHIFT + 8)) | |
| 4179 | +#define PP_DMA_INDEX_BITS ((__builtin_constant_p(PAGE_OFFSET) && \ | |
| 4180 | + PAGE_OFFSET >= PP_DMA_INDEX_MIN_OFFSET && \ | |
| 4181 | + !(PAGE_OFFSET & (PP_DMA_INDEX_MIN_OFFSET - 1))) ? \ | |
| 4182 | + MIN(32, __ffs(PAGE_OFFSET) - PP_DMA_INDEX_SHIFT) : 0) | |
| 4183 | + | |
| 4180 | 4184 | #endif |
| 4181 | 4185 | |
| 4182 | 4186 | #define PP_DMA_INDEX_MASK GENMASK(PP_DMA_INDEX_BITS + PP_DMA_INDEX_SHIFT - 1, \ |