diff for duplicates of <56A8146B.4070601@codeaurora.org> diff --git a/a/1.txt b/N1/1.txt index 9a1fdd9..55826f3 100644 --- a/a/1.txt +++ b/N1/1.txt @@ -77,7 +77,7 @@ space, with assumption device Example1 with ITS page size 64K + device table entry size 8bytes: -Level-1: ((page_order(8MB) ? 1 ) * PAGE_SIZE) /level-one entry size = +Level-1: ((page_order(8MB) – 1 ) * PAGE_SIZE) /level-one entry size = 4M/8 = 19bits Level-2: 64K / 8 (device table entry size) = 13bits. (64K came from default ITS page size) @@ -88,7 +88,7 @@ with 32-Bytes device entry size = handles 30bit DEVID Example2 with ITS page size 16K + device table entry size 8bytes: -Level-1: ((page_order(8MB) ? 1 ) * PAGE_SIZE) /level-one entry size = +Level-1: ((page_order(8MB) – 1 ) * PAGE_SIZE) /level-one entry size = 4M/8 = 19bits Level-2: 16K / 8 (device table entry size) = 11bits. (for ITS hardware that doesn't support 64K page size) @@ -156,9 +156,9 @@ max_ids = level-1 device table size + shift (level-2 device table size) > You can only evaluate the shareability after having written something, > and read it back. Also, the name of the flag is a bit wrong: it is not > the device that needs flushing, bit the ITS Device table. - there is a ?goto retry_baser? statement at the end of ?for loop? that + there is a “goto retry_baser” statement at the end of “for loop” that triggers above code logic that is repeated - whenever the ITS hardware doesn?t stick shareability bit. + whenever the ITS hardware doesn’t stick shareability bit. >> + val |= GITS_BASER_INDIRECT; > Erm... Do you realize that this is an optional feature? What happens if > the HW doesn't support it? @@ -206,7 +206,7 @@ I will rename all references of DEV-ITT to device table. >> + return 0; > It really time we convert this driver to use dma_alloc_coherent. This is > getting out of control... -dma_alloc_coherent() API operates on ?struct device?. We don?t have +dma_alloc_coherent() API operates on “struct device”. We don’t have such device object available in current ITS driver to call DMA API. >> +} >> + @@ -224,8 +224,8 @@ such device object available in current ITS driver to call DMA API. > Under which circumstances can this actually happen? Current driver assumes the device memory has been allocated to cover all possible DEVIDs but is not always true. -This is perfect as long as sparse DEVID is small and if we don?t see the -message ?Device Table too large, reduce its page order to? +This is perfect as long as sparse DEVID is small and if we don’t see the +message “Device Table too large, reduce its page order to” in its_alloc_tables() . As per GICv3-ITS spec, the MAPD command with DEV ID exceeding maximum diff --git a/a/content_digest b/N1/content_digest index e2b49f4..90a3570 100644 --- a/a/content_digest +++ b/N1/content_digest @@ -1,9 +1,17 @@ "ref\01453762723-19477-1-git-send-email-shankerd@codeaurora.org\0" "ref\056A758F4.4010001@arm.com\0" - "From\0shankerd@codeaurora.org (Shanker Donthineni)\0" - "Subject\0[PATCH 1/1] [V2] irqchip: gicv3-its: Introduce indirect device-ITT table support\0" + "From\0Shanker Donthineni <shankerd@codeaurora.org>\0" + "Subject\0Re: [PATCH 1/1] [V2] irqchip: gicv3-its: Introduce indirect device-ITT table support\0" "Date\0Tue, 26 Jan 2016 18:50:51 -0600\0" - "To\0linux-arm-kernel@lists.infradead.org\0" + "To\0Marc Zyngier <marc.zyngier@arm.com>" + linux-kernel@vger.kernel.org + " linux-arm-kernel@lists.infradead.org\0" + "Cc\0Thomas Gleixner <tglx@linutronix.de>" + Jason Cooper <jason@lakedaemon.net> + Vikram Sethi <vikrams@codeaurora.org> + Timur Tabi <timur@codeaurora.org> + Christopher Covington <ccovingt@quicinc.com> + " Tomasz Nowicki <tn@semihalf.com>\0" "\00:1\0" "b\0" "\n" @@ -85,7 +93,7 @@ "\n" "Example1 with ITS page size 64K + device table entry size 8bytes:\n" "\n" - "Level-1: ((page_order(8MB) ? 1 ) * PAGE_SIZE) /level-one entry size = \n" + "Level-1: ((page_order(8MB) \342\200\223 1 ) * PAGE_SIZE) /level-one entry size = \n" "4M/8 = 19bits\n" "Level-2: 64K / 8 (device table entry size) = 13bits. (64K came from \n" "default ITS page size)\n" @@ -96,7 +104,7 @@ "\n" "Example2 with ITS page size 16K + device table entry size 8bytes:\n" "\n" - "Level-1: ((page_order(8MB) ? 1 ) * PAGE_SIZE) /level-one entry size = \n" + "Level-1: ((page_order(8MB) \342\200\223 1 ) * PAGE_SIZE) /level-one entry size = \n" "4M/8 = 19bits\n" "Level-2: 16K / 8 (device table entry size) = 11bits. (for ITS hardware \n" "that doesn't support 64K page size)\n" @@ -164,9 +172,9 @@ "> You can only evaluate the shareability after having written something,\n" "> and read it back. Also, the name of the flag is a bit wrong: it is not\n" "> the device that needs flushing, bit the ITS Device table.\n" - " there is a ?goto retry_baser? statement at the end of ?for loop? that \n" + " there is a \342\200\234goto retry_baser\342\200\235 statement at the end of \342\200\234for loop\342\200\235 that \n" "triggers above code logic that is repeated\n" - " whenever the ITS hardware doesn?t stick shareability bit.\n" + " whenever the ITS hardware doesn\342\200\231t stick shareability bit.\n" ">> +\t\t\tval |= GITS_BASER_INDIRECT;\n" "> Erm... Do you realize that this is an optional feature? What happens if\n" "> the HW doesn't support it?\n" @@ -214,7 +222,7 @@ ">> +\treturn 0;\n" "> It really time we convert this driver to use dma_alloc_coherent. This is\n" "> getting out of control...\n" - "dma_alloc_coherent() API operates on ?struct device?. We don?t have \n" + "dma_alloc_coherent() API operates on \342\200\234struct device\342\200\235. We don\342\200\231t have \n" "such device object available in current ITS driver to call DMA API.\n" ">> +}\n" ">> +\n" @@ -232,8 +240,8 @@ "> Under which circumstances can this actually happen?\n" "Current driver assumes the device memory has been allocated to cover \n" "all possible DEVIDs but is not always true.\n" - "This is perfect as long as sparse DEVID is small and if we don?t see the \n" - "message ?Device Table too large, reduce its page order to?\n" + "This is perfect as long as sparse DEVID is small and if we don\342\200\231t see the \n" + "message \342\200\234Device Table too large, reduce its page order to\342\200\235\n" "in its_alloc_tables() .\n" "\n" "As per GICv3-ITS spec, the MAPD command with DEV ID exceeding maximum \n" @@ -270,4 +278,4 @@ ">\n" "> \tM." -2f834e60170a3af237b773b6203d52269a3224d0929391465dbb8006a6364fe1 +a7d4e70736216287edb2a6f713004c02bd2ffd3ab4f364b5c7ed2d9214ceb701
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.