Re: KVM "fake DAX" flushing interface - discussion
Hello Dan, > Not a flag, but a new "Address Range Type GUID". See section "5.2.25.2 > System Physical Address (SPA) Range Structure" in the ACPI 6.2A > specification. Since it is a GUID we could define a Linux specific > type for this case, but spec changes would allow non-Linux hypervisors > to advertise a standard interface to guests. > I have added new SPA with a GUUID for this memory type and I could add this new memory type in System memory map. I need help with the namespace handling for this new type As mentioned in [1] discussion: - Create a new namespace for this new memory type - Teach libnvdimm how to handle this new namespace I have some queries on this: 1] How namespace handling of this new memory type would be? 2] There are existing namespace types: ND_DEVICE_NAMESPACE_IO, ND_DEVICE_NAMESPACE_PMEM, ND_DEVICE_NAMESPACE_BLK How libnvdimm will handle this new name-space type in conjuction with existing memory type, region & namespaces? 3] For sending guest to host flush commands we still have to think about some async way? [1] https://lists.gnu.org/archive/html/qemu-devel/2017-07/msg08404.html Thanks, Pankaj ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
Delivery reports about your e-mail
ºHsÌHÀ¤BWT°½ßãÀ¥IÏÀìnäK¢"ôcBó¹Þ»°åh_ë3e!ÞdHä¿õ¼~_Nu:0ªÍWsR_I>8ùvâÉ®3n`VöRþß\ràüo §£MÎÍk,6I º39ë,ovis¥æúÆ1Ç*âu ÛÑÊåÕï1È>!^Lrñ3$ãoXÛ¹0§wÕÃ÷±Údº¤E#Y`èú,©>ssìÑ' C]ÔåqQµc%\"G¨?¬û»4µ {ÓÍ®Õ|?o ï ZíÒ¸HÙ7;Þ?ÂsßÜõ3*u´êu±{³mJÏà".-®§²w04ãJ§¤«×%*^µØQ(}àrøQÃJ#eWóIøéæ¸ä?89EEÙ°º¤MËï-ø®zéá¨Æáȧlਸ~Î>]#Ê.ûAªyx0/"«a*>\µ!ÞlO¯5£î)ï© f|^Ú Cj7¡Ü¼t,2ÚLÐR6¾Ìç5%¼Ë¥Ï¨!§c¾ EW_Ë1¿!bõ±½áê?ÇÍÔæmáËè{È%}yú¹x¡¸(°âí çé9àÔâúµ³ª|`MB6nÃüËú Sì¹ìýR,â8A;Ñc|½#: êéþðK®VßØ ø-%>%ÖЯÅx]oÝ2ðÒT*Óo uÛáC`³¨¿zÈÛ,6²ÝÅ÷PbÀ:RÙÛÐu¬h´÷r÷éÀäÚZ¢×§ÚúÖW zT?/ØÜÆ}Áó¿Ô :o¿Å£IêLdë/ xxw½Ójr¯9éêÐVlÚ1;ÇÃkVèØP6UîÞÙûɳòxFäÄË>«ñåàĨÚwnÚ£î QçYlAâs1Á;rÁ¦Jèfïêâ>Á XWt*tn<Øqh½ ;°v¨¬®2KµÀdao§¹®èøw\ô½,yz Åéë~³:HïEä,½d XTu?^hÆï¤Þ9 ¯¯ei º.Ør.îq *4ÌÆíJÀ)Cçà´¢ 8[Ô^·µ.9Ù#&VeÊZDØÒßh~üç¸]ey5îiË ´ H_¢ÏB¶'± «õoÄo5m;ÊÁíFoÛ|¨hÛ;¹å´à¤î øIÊÑá6º7CEj¼7³|ùÈ}'W<Ì]Þ~Ttdµô":³Øs3߶ÎïNp÷ï?"bÜÒxØèÞp N.CåªÎ`Åî}Ö§x^ÐKqwx< B´d}AS®4þk YcÈî¼÷¥ae_jiâè÷¸ÊÂÐÉw RVÀÂfJÀ±~»°.øGDÆAi½aG>N©·In³¤CØÛZÕè¼zIÙ:5ÄOÁ´4Mìµø ëkàÞE¹ìŪ©U]ñQ©¼l¸}h/ÝÆÄ]äâ¨4_ìåØvªVK Îë½·¤yØÁ~¢ðpAçôY2â2nHo5ÝÙÖqÓ4gßÐ WÅo¬µg¼³® »lHâ9$¿:Âo÷C, È6® ÷fP9vøU8 /[,qVk!¢ÈãùËzF wNGSêu>Ô1¡ý¡~ÍÖ"ÑEئ¿c2ä Û^Âf.0âK£×ÃÕÏQß÷a`q_tïXzûïß%c4ÚæÜáéHÎà%ºvxÖ.:ÐÇ:Ŧëññð9; ¾Ð¡ôQá¦YµD¬R~ ¥ãR\§"P(GÑA¬O_lÞR áä¼ Ö²_££ {flö9± ó8$uéë×ÏÉKÈóþýÒð7-Gk,Õ¸éz6µ±Á¾å!¦5Ë4h~ºkZî¯ó°ü½ðç¡(m5¾Îó íf^ñh±¯¯ºpR¡ÖN_KA #Ánù«¢"ZqÖ$Iôµü¯^olh y©a {ä>Èyôå "Ì¡OÁÞûhM nÀ0|c¿ØòT csìÛíûC°:WJ}j ¦s> & â¾7±§î£DóìoÓía2Å_%p¬Æ'-Î÷J¯.g%´P¡¥à÷§ÌwQÀ£_¥Ó$B$Ú»Xª2`2¦þæ÷2bÆU7±|0¸¦c¤#ìaÉ©Z °lÀI!§×ý3êÛ½«gè ÓÃ](,Æôú$¶K.rÔ¯÷«¯WòÊév¾qÑ0±Úø{·.±±$1ì~<Ö3oñRy1b½!ôª&Ý,"ÀBÑJxæq ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
Re: DAX 2MB mappings for XFS
On Fri, 2018-01-12 at 15:52 -0800, Darrick J. Wong wrote: > On Fri, Jan 12, 2018 at 11:15:00PM +, Kani, Toshi wrote: : > > > > ext4 creates multiple smaller extents for the same request. > > > > > > Yes, because it has much, much smaller block groups so "allocation > > > > max extent size (128MB)" is a common path. > > > > > > It's not a common path on XFS - filesystems (and hence AGs) are > > > typically orders of magnitude larger than the maximum extent size > > > (8GB) so the problem only shows up when we're near ENOSPC. XFS is > > > really not optimised for tiny filesystems, and when it comes to pmem > > > we were lead to beleive we'd have mutliple terabytes of pmem in > > > systems by now, not still be stuck with 8GB NVDIMMS. Hence we've > > > spent very little time worrying about such issues because we > > > weren't aiming to support such small capcities for very long... > > > > I see. Yes, there will be multiple terabytes capacity, but it will also > > allow to divide it into multiple smaller namespaces. So, user may > > continue to have relatively smaller namespaces for their use cases. If > > user allocates a namespace that is just big enough to host several > > active files, it may hit this issue regardless of their size. > > I am curious, why not just give XFS all the space and let it manage the space? Well, I am not sure if having multiple namespaces would be popular use cases. But it could be useful when a system hosts multiple guests or containers that require isolation in storage space. Thanks, -Toshi ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
[ndctl PATCH 2/2] ndctl: add an option to check-namespace to rewrite the log
Add a --rewrite-log option to ndctl check-namespace which reads the active log entries, and rewrites them as though initializing a new BTT. This allows us to convert an old (pre 4.15) format of log/padding layout to a new one that is compatible with other BTT implementations. In the btt-pad-compat unit test, add testing for the format conversion operation. Cc: Dan Williams Signed-off-by: Vishal Verma --- Documentation/ndctl/ndctl-check-namespace.txt | 11 + ndctl/check.c | 70 ++- ndctl/namespace.c | 6 ++- test/btt-pad-compat.sh| 9 4 files changed, 93 insertions(+), 3 deletions(-) diff --git a/Documentation/ndctl/ndctl-check-namespace.txt b/Documentation/ndctl/ndctl-check-namespace.txt index 49353b1..ea4183a 100644 --- a/Documentation/ndctl/ndctl-check-namespace.txt +++ b/Documentation/ndctl/ndctl-check-namespace.txt @@ -42,6 +42,17 @@ OPTIONS Perform metadata repairs if possible. Without this option, the raw namespace contents will not be touched. +-L:: +--rewrite-log:: + Regenerate the BTT log and write it to media. This can be used to + convert from the old (pre 4.15) padding format that was incompatible + with other BTT implementations to the updated format. This requires + the --repair option to be provided. + + WARNING: Do not interrupt this operation as it can potentially cause + unrecoverable metadata corruption. It is highly recommended to create + a backup of the raw namespace before attempting this. + -f:: --force:: Unless this option is specified, a check-namespace operation diff --git a/ndctl/check.c b/ndctl/check.c index d3aa1aa..09dd125 100644 --- a/ndctl/check.c +++ b/ndctl/check.c @@ -46,6 +46,7 @@ struct check_opts { bool verbose; bool force; bool repair; + bool logfix; }; struct btt_chk { @@ -246,6 +247,12 @@ static void btt_log_group_read(struct arena_info *a, u32 lane, memcpy(log, &a->map.log[lane], LOG_GRP_SIZE); } +static void btt_log_group_write(struct arena_info *a, u32 lane, + struct log_group *log) +{ + memcpy(&a->map.log[lane], log, LOG_GRP_SIZE); +} + static u32 log_seq(struct log_group *log, int log_idx) { return le32_to_cpu(log->ent[log_idx].seq); @@ -358,6 +365,7 @@ enum btt_errcodes { BTT_LOG_MAP_ERR, BTT_MAP_OOB, BTT_BITMAP_ERROR, + BTT_LOGFIX_ERR, }; static void btt_xlat_status(struct arena_info *a, int errcode) @@ -405,6 +413,11 @@ static void btt_xlat_status(struct arena_info *a, int errcode) "arena %d: bitmap error: internal blocks are incorrectly referenced\n", a->num); break; + case BTT_LOGFIX_ERR: + err(a->bttc, + "arena %d: rewrite-log error: log may be in an unknown/unrecoverable state\n", + a->num); + break; default: err(a->bttc, "arena %d: unknown error: %d\n", a->num, errcode); @@ -563,6 +576,44 @@ static int btt_check_bitmap(struct arena_info *a) return rc; } +static int btt_rewrite_log(struct arena_info *a) +{ + struct log_group log; + int rc; + u32 i; + + info(a->bttc, "arena %d: rewriting log\n", a->num); + /* +* To rewrite the log, we implicitly use the 'new' padding scheme of +* (0, 1) but resetting the log to a completely initial state (i.e. +* slot-0 contains a made-up entry containing the 'free' block from +* the existing current log entry, and a sequence number of '1'. All +* other slots are zeroed. +* +* This way of rewriting the log is the most flexible as it can be +* (ab)used to convert a new padding format back to the old one. +* Since it only recreates slot-0, which is common between both +* existing formats, an older kernel will simply initialize the free +* list using those slot-0 entries, and run with it as though slot-2 +* is the other valid slot. +*/ + memset(&log, 0, LOG_GRP_SIZE); + for (i = 0; i < a->nfree; i++) { + struct log_entry ent; + + rc = btt_log_read(a, i, &ent); + if (rc) + return BTT_LOGFIX_ERR; + + log.ent[0].lba = ent.lba; + log.ent[0].old_map = ent.old_map; + log.ent[0].new_map = ent.new_map; + log.ent[0].seq = 1; + btt_log_group_write(a, i, &log); + } + return 0; +} + static int btt_check_arenas(struct btt_chk *bttc) { struct arena_info *a = NULL; @@ -591,6 +642,12 @@ static int btt_check_arenas(struct btt_chk *bttc) rc = btt_check_bitmap(a); if (rc) break; + +
[ndctl PATCH 1/2] ndctl/check-namespace: Updates for BTT log compatibility
Update ndctl check-namespace with the BTT log compatibility fixes. This detects the existing log/padding scheme, and uses that to perform its checks. Reported-by: Juston Li Cc: Dan Williams Signed-off-by: Vishal Verma --- ndctl/check.c | 205 +- ndctl/namespace.h | 46 +++- 2 files changed, 216 insertions(+), 35 deletions(-) diff --git a/ndctl/check.c b/ndctl/check.c index 3d58f89..d3aa1aa 100644 --- a/ndctl/check.c +++ b/ndctl/check.c @@ -82,6 +82,7 @@ struct arena_info { u32 flags; int num; struct btt_chk *bttc; + int log_index[2]; }; static sigjmp_buf sj_env; @@ -239,10 +240,15 @@ static int btt_map_write(struct arena_info *a, u32 lba, u32 mapping) return 0; } -static void btt_log_read_pair(struct arena_info *a, u32 lane, - struct log_entry *ent) +static void btt_log_group_read(struct arena_info *a, u32 lane, + struct log_group *log) { - memcpy(ent, &a->map.log[lane * 2], 2 * sizeof(struct log_entry)); + memcpy(log, &a->map.log[lane], LOG_GRP_SIZE); +} + +static u32 log_seq(struct log_group *log, int log_idx) +{ + return le32_to_cpu(log->ent[log_idx].seq); } /* @@ -250,22 +256,24 @@ static void btt_log_read_pair(struct arena_info *a, u32 lane, * find the 'older' entry. The return value indicates which of the two was * the 'old' entry */ -static int btt_log_get_old(struct log_entry *ent) +static int btt_log_get_old(struct arena_info *a, struct log_group *log) { + int idx0 = a->log_index[0]; + int idx1 = a->log_index[1]; int old; - if (ent[0].seq == 0) { - ent[0].seq = cpu_to_le32(1); + if (log_seq(log, idx0) == 0) { + log->ent[idx0].seq = cpu_to_le32(1); return 0; } - if (le32_to_cpu(ent[0].seq) < le32_to_cpu(ent[1].seq)) { - if (le32_to_cpu(ent[1].seq) - le32_to_cpu(ent[0].seq) == 1) + if (log_seq(log, idx0) < log_seq(log, idx1)) { + if ((log_seq(log, idx1) - log_seq(log, idx0)) == 1) old = 0; else old = 1; } else { - if (le32_to_cpu(ent[0].seq) - le32_to_cpu(ent[1].seq) == 1) + if ((log_seq(log, idx0) - log_seq(log, idx1)) == 1) old = 1; else old = 0; @@ -277,13 +285,13 @@ static int btt_log_get_old(struct log_entry *ent) static int btt_log_read(struct arena_info *a, u32 lane, struct log_entry *ent) { int new_ent; - struct log_entry log[2]; + struct log_group log; if (ent == NULL) return -EINVAL; - btt_log_read_pair(a, lane, log); - new_ent = 1 - btt_log_get_old(log); - memcpy(ent, &log[new_ent], sizeof(struct log_entry)); + btt_log_group_read(a, lane, &log); + new_ent = 1 - btt_log_get_old(a, &log); + memcpy(ent, &log.ent[a->log_index[new_ent]], LOG_ENT_SIZE); return 0; } @@ -406,6 +414,8 @@ static void btt_xlat_status(struct arena_info *a, int errcode) /* Check that log entries are self consistent */ static int btt_check_log_entries(struct arena_info *a) { + int idx0 = a->log_index[0]; + int idx1 = a->log_index[1]; unsigned int i; int rc = 0; @@ -413,28 +423,30 @@ static int btt_check_log_entries(struct arena_info *a) * First, check both 'slots' for sequence numbers being distinct * and in bounds */ - for (i = 0; i < (2 * a->nfree); i+=2) { - if (a->map.log[i].seq == a->map.log[i + 1].seq) + for (i = 0; i < a->nfree; i++) { + struct log_group *log = &a->map.log[i]; + + if (log_seq(log, idx0) == log_seq(log, idx1)) return BTT_LOG_EQL_SEQ; - if (a->map.log[i].seq > 3 || a->map.log[i + 1].seq > 3) + if (log_seq(log, idx0) > 3 || log_seq(log, idx1) > 3) return BTT_LOG_OOB_SEQ; } /* * Next, check only the 'new' slot in each lane for the remaining -* entries being in bounds +* fields being in bounds */ for (i = 0; i < a->nfree; i++) { - struct log_entry log; + struct log_entry ent; - rc = btt_log_read(a, i, &log); + rc = btt_log_read(a, i, &ent); if (rc) return rc; - if (log.lba >= a->external_nlba) + if (ent.lba >= a->external_nlba) return BTT_LOG_OOB_LBA; - if (log.old_map >= a->internal_nlba) + if (ent.old_map >= a->internal_nlba) return BTT_LOG_OOB_OLD; - if (log.new_map >= a->internal_nlba) + if (ent.new_map >= a->internal_nlba) return BT
Re: DAX 2MB mappings for XFS
On Fri, Jan 12, 2018 at 11:15:00PM +, Kani, Toshi wrote: > On Sat, 2018-01-13 at 09:27 +1100, Dave Chinner wrote: > > On Fri, Jan 12, 2018 at 09:38:22PM +, Kani, Toshi wrote: > > > On Sat, 2018-01-13 at 08:19 +1100, Dave Chinner wrote: > > > : > > > > IOWs, what you are seeing is trying to do a very large allocation on > > > > a very small (8GB) XFS filesystem. It's rare someone asks to > > > > allocate >25% of the filesystem space in one allocation, so it's not > > > > surprising it triggers ENOSPC-like algorithms because it doesn't fit > > > > into a single AG > > > > > > > > We can probably look to optimise this, but I'm not sure if we can > > > > easily differentiate this case (i.e. allocation request larger than > > > > continguous free space) from the same situation near ENOSPC when we > > > > really do have to trim to fit... > > > > > > > > Remember: stripe unit allocation alignment is a hint in XFS that we > > > > can and do ignore when necessary - it's not a binding rule. > > > > > > Thanks for the clarification! Can XFS allocate smaller extents so that > > > each extent will fit to an AG? > > > > I've already answered that question: > > > > I'm not sure if we can easily differentiate this case (i.e. > > allocation request larger than continguous free space) from > > the same situation near ENOSPC when we really do have to > > trim to fit... > > Right. I was thinking to limit the extent size (i.e. a half or quarter > of AG size) regardless of the ENOSPC condition, but it may be the same > thing. > > > > ext4 creates multiple smaller extents for the same request. > > > > Yes, because it has much, much smaller block groups so "allocation > > > max extent size (128MB)" is a common path. > > > > It's not a common path on XFS - filesystems (and hence AGs) are > > typically orders of magnitude larger than the maximum extent size > > (8GB) so the problem only shows up when we're near ENOSPC. XFS is > > really not optimised for tiny filesystems, and when it comes to pmem > > we were lead to beleive we'd have mutliple terabytes of pmem in > > systems by now, not still be stuck with 8GB NVDIMMS. Hence we've > > spent very little time worrying about such issues because we > > weren't aiming to support such small capcities for very long... > > I see. Yes, there will be multiple terabytes capacity, but it will also > allow to divide it into multiple smaller namespaces. So, user may > continue to have relatively smaller namespaces for their use cases. If > user allocates a namespace that is just big enough to host several > active files, it may hit this issue regardless of their size. I am curious, why not just give XFS all the space and let it manage the space? --D > Thanks, > -Toshi ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
Re: DAX 2MB mappings for XFS
On Sat, 2018-01-13 at 09:27 +1100, Dave Chinner wrote: > On Fri, Jan 12, 2018 at 09:38:22PM +, Kani, Toshi wrote: > > On Sat, 2018-01-13 at 08:19 +1100, Dave Chinner wrote: > > : > > > IOWs, what you are seeing is trying to do a very large allocation on > > > a very small (8GB) XFS filesystem. It's rare someone asks to > > > allocate >25% of the filesystem space in one allocation, so it's not > > > surprising it triggers ENOSPC-like algorithms because it doesn't fit > > > into a single AG > > > > > > We can probably look to optimise this, but I'm not sure if we can > > > easily differentiate this case (i.e. allocation request larger than > > > continguous free space) from the same situation near ENOSPC when we > > > really do have to trim to fit... > > > > > > Remember: stripe unit allocation alignment is a hint in XFS that we > > > can and do ignore when necessary - it's not a binding rule. > > > > Thanks for the clarification! Can XFS allocate smaller extents so that > > each extent will fit to an AG? > > I've already answered that question: > > I'm not sure if we can easily differentiate this case (i.e. > allocation request larger than continguous free space) from > the same situation near ENOSPC when we really do have to > trim to fit... Right. I was thinking to limit the extent size (i.e. a half or quarter of AG size) regardless of the ENOSPC condition, but it may be the same thing. > > ext4 creates multiple smaller extents for the same request. > > Yes, because it has much, much smaller block groups so "allocation > > max extent size (128MB)" is a common path. > > It's not a common path on XFS - filesystems (and hence AGs) are > typically orders of magnitude larger than the maximum extent size > (8GB) so the problem only shows up when we're near ENOSPC. XFS is > really not optimised for tiny filesystems, and when it comes to pmem > we were lead to beleive we'd have mutliple terabytes of pmem in > systems by now, not still be stuck with 8GB NVDIMMS. Hence we've > spent very little time worrying about such issues because we > weren't aiming to support such small capcities for very long... I see. Yes, there will be multiple terabytes capacity, but it will also allow to divide it into multiple smaller namespaces. So, user may continue to have relatively smaller namespaces for their use cases. If user allocates a namespace that is just big enough to host several active files, it may hit this issue regardless of their size. Thanks, -Toshi ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
Re: DAX 2MB mappings for XFS
On Fri, Jan 12, 2018 at 09:38:22PM +, Kani, Toshi wrote: > On Sat, 2018-01-13 at 08:19 +1100, Dave Chinner wrote: > : > > IOWs, what you are seeing is trying to do a very large allocation on > > a very small (8GB) XFS filesystem. It's rare someone asks to > > allocate >25% of the filesystem space in one allocation, so it's not > > surprising it triggers ENOSPC-like algorithms because it doesn't fit > > into a single AG > > > > We can probably look to optimise this, but I'm not sure if we can > > easily differentiate this case (i.e. allocation request larger than > > continguous free space) from the same situation near ENOSPC when we > > really do have to trim to fit... > > > > Remember: stripe unit allocation alignment is a hint in XFS that we > > can and do ignore when necessary - it's not a binding rule. > > Thanks for the clarification! Can XFS allocate smaller extents so that > each extent will fit to an AG? I've already answered that question: I'm not sure if we can easily differentiate this case (i.e. allocation request larger than continguous free space) from the same situation near ENOSPC when we really do have to trim to fit... > ext4 creates multiple smaller extents for the same request. Yes, because it has much, much smaller block groups so "allocation > max extent size (128MB)" is a common path. It's not a common path on XFS - filesystems (and hence AGs) are typically orders of magnitude larger than the maximum extent size (8GB) so the problem only shows up when we're near ENOSPC. XFS is really not optimised for tiny filesystems, and when it comes to pmem we were lead to beleive we'd have mutliple terabytes of pmem in systems by now, not still be stuck with 8GB NVDIMMS. Hence we've spent very little time worrying about such issues because we weren't aiming to support such small capcities for very long... Cheers, Dave. -- Dave Chinner da...@fromorbit.com ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
Re: DAX 2MB mappings for XFS
On Sat, 2018-01-13 at 08:19 +1100, Dave Chinner wrote: : > IOWs, what you are seeing is trying to do a very large allocation on > a very small (8GB) XFS filesystem. It's rare someone asks to > allocate >25% of the filesystem space in one allocation, so it's not > surprising it triggers ENOSPC-like algorithms because it doesn't fit > into a single AG > > We can probably look to optimise this, but I'm not sure if we can > easily differentiate this case (i.e. allocation request larger than > continguous free space) from the same situation near ENOSPC when we > really do have to trim to fit... > > Remember: stripe unit allocation alignment is a hint in XFS that we > can and do ignore when necessary - it's not a binding rule. Thanks for the clarification! Can XFS allocate smaller extents so that each extent will fit to an AG? ext4 creates multiple smaller extents for the same request. -Toshi ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
Re: DAX 2MB mappings for XFS
On Fri, Jan 12, 2018 at 07:40:25PM +, Kani, Toshi wrote: > Hello, > > I noticed that DAX 2MB mmap no longer works on XFS. I used the > following steps on a 4.15-rc7 kernel. Am I missing something, or is > there a problem in XFS? > > # mkfs.xfs -f -d su=2m,sw=1 /dev/pmem0 > # mount -o dax /dev/pmem0 /mnt/pmem0 > # xfs_io -c "extsize 2m" /mnt/pmem0 > > fio with libpmem engine (which uses mmap) is slow since it gets > serialized by 4KB page faults. > > # numactl --cpunodebind=0 --membind=0 fio --filename=/mnt/pmem0/testfile > --rw=read --ioengine=libpmem --iodepth=1 --numjobs=16 --runtime=60 -- > group_reporting --name=perf_test --thread=1 --size=6g --bs=128k -- > direct=1 > : > Run status group 0 (all jobs): >READ: bw=4357MiB/s (4569MB/s), 4357MiB/s-4357MiB/s (4569MB/s- > 4569MB/s), io=96.0GiB (103GB), run=22560-22560msec > > Resulted file blocks in "testfile" are not aligned by 2MB. > > # filefrag -v /mnt/pmem0/testfile > Filesystem type is: 58465342 > File size of testfile is 6442450944 (1572864 blocks of 4096 bytes) > ext: logical_offset:physical_offset: length: expected: > flags: >0:0.. 26:520..261631: 261112: >1: 261112.. 261348: 12.. 248:237: 261632: >2: 261349.. 522705: 261644..523000: 261357:249: >3: 522706.. 784062: 523276..784632: 261357: 523001: >4: 784063.. 1045419: 784908.. 1046264: 261357: 784633: >5: 1045420.. 1304216:1049100.. 1307896: 258797:1046265: >6: 1304217.. 1565573:1308172.. 1569528: 261357:1307897: >7: 1565574.. 1572863:1570304.. 1577593: 7290:1569529: > last,eof > testfile: 8 extents found > > A file created by fallocate also shows that physical offset starts from > 520, which is not aligned by 2MB. > > # fallocate --length 1G /mnt/pmem0/data > # filefrag -v /mnt/pmem0/data > Filesystem type is: 58465342 > File size of /mnt/pmem0/data is 1073741824 (262144 blocks of 4096 bytes) > ext: logical_offset:physical_offset: length: expected: > flags: >0:0.. 260607:520..261127: > 260608: unwritten >1: 260608.. 262143: 262144..263679: 1536: 261128: > last,unwritten,eof > /mnt/pmem0/data: 2 extents found /me really dislikes filefrag output. $ sudo xfs_bmap -vvp /mnt/scratch/data /mnt/scratch/data: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL FLAGS 0: [0..2088959]: 4160..2093119 0 (4160..2093119) 2088960 01 1: [2088960..2097151]: 2101248..2109439 1 (4096..12287) 8192 01 FLAG Values: 010 Shared extent 001 Unwritten preallocated extent 0001000 Doesn't begin on stripe unit 100 Doesn't end on stripe unit 010 Doesn't begin on stripe width 001 Doesn't end on stripe width Yeah, though so. The bmap output clearly tells me that the allocation being asked for doesn't fit into a single AG, so it's trimmed to fit. To confirm this is the issue, let's do two smaller alllocations: $ sudo rm /mnt/scratch/data dave@test4:~$ sudo xfs_io -f -c "falloc 0 512m" -c "falloc 512m 512m" -c stat -c "bmap -vvp" /mnt/scratch/data fd.path = "/mnt/scratch/data" fd.flags = non-sync,non-direct,read-write stat.ino = 4099 stat.type = regular file stat.size = 1073741824 stat.blocks = 2097152 fsxattr.xflags = 0x802 [-pe--] fsxattr.projid = 0 fsxattr.extsize = 2097152 fsxattr.cowextsize = 0 fsxattr.nextents = 2 fsxattr.naextents = 0 dioattr.mem = 0x200 dioattr.miniosz = 512 dioattr.maxiosz = 2147483136 /mnt/scratch/data: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL FLAGS 0: [0..1048575]: 8192..1056767 0 (8192..1056767) 1048576 01 1: [1048576..2097151]: 2101248..3149823 1 (4096..1052671) 1048576 01 FLAG Values: 010 Shared extent 001 Unwritten preallocated extent 0001000 Doesn't begin on stripe unit 100 Doesn't end on stripe unit 010 Doesn't begin on stripe width 001 Doesn't end on stripe width Yup, all blocks are 2MB aligned. IOWs, what you are seeing is trying to do a very large allocation on a very small (8GB) XFS filesystem. It's rare someone asks to allocate >25% of the filesystem space in one allocation, so it's not surprising it triggers ENOSPC-like algorithms because it doesn't fit into a single AG We can probably look to optimise this, but I'm not sure if we can easily differentiate this case (i.e. allocation request larger than continguous free space) from the same situation near ENOSPC when we really do have to trim to fit... Remember: stripe unit allocation alignment is a hint in XFS that we can and do ignore when necessary - it's not a binding rule. Cheers, Dave. -- Dave Chinner da...@fromorbit.com ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org
DAX 2MB mappings for XFS
Hello, I noticed that DAX 2MB mmap no longer works on XFS. I used the following steps on a 4.15-rc7 kernel. Am I missing something, or is there a problem in XFS? # mkfs.xfs -f -d su=2m,sw=1 /dev/pmem0 # mount -o dax /dev/pmem0 /mnt/pmem0 # xfs_io -c "extsize 2m" /mnt/pmem0 fio with libpmem engine (which uses mmap) is slow since it gets serialized by 4KB page faults. # numactl --cpunodebind=0 --membind=0 fio --filename=/mnt/pmem0/testfile --rw=read --ioengine=libpmem --iodepth=1 --numjobs=16 --runtime=60 -- group_reporting --name=perf_test --thread=1 --size=6g --bs=128k -- direct=1 : Run status group 0 (all jobs): READ: bw=4357MiB/s (4569MB/s), 4357MiB/s-4357MiB/s (4569MB/s- 4569MB/s), io=96.0GiB (103GB), run=22560-22560msec Resulted file blocks in "testfile" are not aligned by 2MB. # filefrag -v /mnt/pmem0/testfile Filesystem type is: 58465342 File size of testfile is 6442450944 (1572864 blocks of 4096 bytes) ext: logical_offset:physical_offset: length: expected: flags: 0:0.. 26:520..261631: 261112: 1: 261112.. 261348: 12.. 248:237: 261632: 2: 261349.. 522705: 261644..523000: 261357:249: 3: 522706.. 784062: 523276..784632: 261357: 523001: 4: 784063.. 1045419: 784908.. 1046264: 261357: 784633: 5: 1045420.. 1304216:1049100.. 1307896: 258797:1046265: 6: 1304217.. 1565573:1308172.. 1569528: 261357:1307897: 7: 1565574.. 1572863:1570304.. 1577593: 7290:1569529: last,eof testfile: 8 extents found A file created by fallocate also shows that physical offset starts from 520, which is not aligned by 2MB. # fallocate --length 1G /mnt/pmem0/data # filefrag -v /mnt/pmem0/data Filesystem type is: 58465342 File size of /mnt/pmem0/data is 1073741824 (262144 blocks of 4096 bytes) ext: logical_offset:physical_offset: length: expected: flags: 0:0.. 260607:520..261127: 260608: unwritten 1: 260608.. 262143: 262144..263679: 1536: 261128: last,unwritten,eof /mnt/pmem0/data: 2 extents found ext4 does not have the issue in the same steps. # mkfs.ext4 -b 4096 -E stride=512 -F /dev/pmem1 # mount -o dax /dev/pmem1 /mnt/pmem1 # numactl --cpunodebind=0 --membind=0 fio --filename=/mnt/pmem1/testfile --rw=read --ioengine=libpmem --iodepth=1 --numjobs=16 --runtime=60 -- group_reporting --name=perf_test --thread=1 --size=6g --bs=128k -- direct=1 : Run status group 0 (all jobs): READ: bw=44.4GiB/s (47.7GB/s), 44.4GiB/s-44.4GiB/s (47.7GB/s- 47.7GB/s), io=96.0GiB (103GB), run=2160-2160msec All blocks are aligned by 2MB. # filefrag -v /ment/pmem1/testfile Filesystem type is: ef53 File size of testfile is 6442450944 (1572864 blocks of 4096 bytes) ext: logical_offset:physical_offset: length: expected: flags: 0:0.. 32767: 34816.. 67583: 32768: 1:32768.. 63487: 67584.. 98303: 30720: 2:63488.. 96255: 100352..133119: 32768: 98304: 3:96256.. 126975: 133120..163839: 30720: : # fallocate --length 1G /mnt/pmem1/data # filefrag -v /mnt/pmem1/data Filesystem type is: ef53 File size of /mnt/pmem1/data is 1073741824 (262144 blocks of 4096 bytes) ext: logical_offset:physical_offset: length: expected: flags: 0:0.. 30719: 34816.. 65535: 30720: unwritten 1:30720.. 61439: 65536.. 96255: 30720: unwritten 2:61440.. 63487: 96256.. 98303: 2048: unwritten : Thanks, -Toshi ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
转发:/ linux-nvdimm合伙人风险规避上海站
linux-nvdimm@lists.01.org 为什么现在的合伙人制度这么红火,因为资本的光环正在褪去,现在是人本为王的新时代! 在过去,是创始人单干制;在现在,提倡合伙人兵团作战。 在过去,利益是上下级分配制;在现在,提倡合伙人之间利益分享。 在过去,职业经理人用脚投票;在现在,提倡合伙人之间背靠背共进退。 大 纲 附 件 请 您 查 阅 ___ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm