Message ID | 1623915111-43630-2-git-send-email-liweihang@huawei.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | RDMA/hns: Use new interfaces to write/read fields | expand |
On Thu, Jun 17, 2021 at 03:31:44PM +0800, Weihang Li wrote: > From: Lang Cheng <chenglang@huawei.com> > > Fix complains from sparse about "dubious: x & !y" when calling > hr_reg_write(ctx, field, !!val). Where is this from? I'm not convinced you should have the temporary here given how much magics are involved in this stuff that rely on builtin_constant_p Jason
On 2021/6/18 6:41, Jason Gunthorpe wrote: > On Thu, Jun 17, 2021 at 03:31:44PM +0800, Weihang Li wrote: >> From: Lang Cheng <chenglang@huawei.com> >> >> Fix complains from sparse about "dubious: x & !y" when calling >> hr_reg_write(ctx, field, !!val). > > Where is this from? > > I'm not convinced you should have the temporary here given how much > magics are involved in this stuff that rely on builtin_constant_p > > Jason > The warning comes from: #define __BF_FIELD_CHECK(_mask, _reg, _val, _pfx) \ ... BUILD_BUG_ON_MSG(__builtin_constant_p(_val) ? \ ~((_mask) >> __bf_shf(_mask)) & (_val) : 0, \ _pfx "value too large for the field"); \ I will change 'hr_reg_write(ctx, field, !!val)' to 'hr_reg_write(ctx, field, val ? 1 : 0)'. The latter is not as succinct as the former, but can avoid the sparse warning. Thanks Weihang
diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h index 3a5658f..eb2d670 100644 --- a/drivers/infiniband/hw/hns/hns_roce_common.h +++ b/drivers/infiniband/hw/hns/hns_roce_common.h @@ -79,9 +79,10 @@ #define _hr_reg_write(ptr, field_type, field_h, field_l, val) \ ({ \ + u32 _val = val; \ _hr_reg_clear(ptr, field_type, field_h, field_l); \ *((__le32 *)ptr + (field_h) / 32) |= cpu_to_le32(FIELD_PREP( \ - GENMASK((field_h) % 32, (field_l) % 32), val)); \ + GENMASK((field_h) % 32, (field_l) % 32), _val)); \ }) #define hr_reg_write(ptr, field, val) _hr_reg_write(ptr, field, val)